AI blog APT blog Attack blog BigBrother blog BotNet blog CyberCrime blog Cyber blog Cryptocurrency blog Exploit blog Hacking blog ICS blog Incident blog IoT blog Malware blog OS Blog Phishing blog Ransom blog Safety blog Security blog Social blog Spam blog Vulnerebility blog
DATE | NAME | Info | CATEG. | WEB |
|
10.10.26 |
Runtime AI Defense in a shared responsibility model | This last February, by their own account, an employee at Meta connected an open source agent to their inbox and told it to suggest which emails to archive or delete, and to confirm with them before acting. Despite the guardrails of the prompt, the agent started deleting. Their stop commands didn't work and they had to manually kill the process. | AI blog | Sysdig |
|
10.10.26 |
Building an evidence-grounded agentic security operations harness on Cloudflare | Security alerts rarely arrive one at a time. A single alert can cause a spike across the environment, requiring a human analyst to decide which alerts are related and what they mean. When multiple arrive at the same time, it can quickly overwhelm even a seasoned security analyst. Enter the alert paradox. Now, our built-in, multi-AI-agent security operations harness can handle more of this work at Cloudflare scale. | AI blog | Cloudflare Blog |
|
10.10.26 |
Meet AI Detection Engineer, the Newest Member of Your SOC | AI Detection Engineer investigates, builds, tests, and deploys a tuned detection for confirmed misjudgments your team reports, then shows the verifiable impact. | AI blog | Abnormal AI |
|
10.10.26 |
AI and the Evolving Threat Landscape: What Public Sector CISOs Need to Know Now | Public sector CISOs need to grapple with challenges brought about by AI threats: up to 3,600 AI CVEs in 2026, autonomous ransomware, and critical infrastructure risk. | AI blog | TrendAI (US) |
|
10.10.26 |
StepSecurity Now Inventories AI Agent Skills in Your GitHub Repositories and on Developer Machines | AI agent skills live in two places. Developers install them on their own machines, and teams commit them to repositories so everyone who works in the codebase gets them. A skill often travels between the two: a developer installs it, finds it useful, and commits it; everyone who clones the repository then has it. | AI blog | StepSecurity |
|
10.10.26 |
Leaked AI & MCP Credentials: What They Expose | One record in the corpus carries the argument before we even start counting. It’s a project configuration file for an AI coding assistant that wires several named tool servers behind a single agent. | AI blog | FLARE.IO |
|
10.10.26 |
AI SOC automation: Inbox zero for your alert queue | AlertZero brings AI SOC automation to Elastic Security with four agents, one job each, so the queue stops setting your priorities. Nothing changes in your environment without your approval. | AI blog | Elastic Security Labs |
|
10.10.26 |
Semantic Privilege Escalation: The AI Security Threat Your Access Controls Can't See | In 1988, Norm Hardy published a short paper about an incident at Tymshare roughly eleven years earlier. A Fortran compiler lived in a system directory called SYSX, and so it could record usage statistics there, administrators had granted it a "home files license" that let it write any file in that directory. | AI blog | Proofpoint US |
|
10.10.26 |
Criminal AI, powered by OpenRouter | Kriminal, DadGPT, and WormGPT are three of the most recognizable tools in the criminal AI marketplace today. Kriminal and DadGPT sell themselves as unrestricted AI: no filters, no refusals, built specifically to say “yes” where a mainstream model would say “no.” WormGPT takes the opposite approach, marketing itself as an “Ethical Hacking AI” for “cybersecurity professionals” — camouflage that lets it hide in plain sight. | AI blog | THREATDOWN |
|
10.10.26 |
How Attackers Can Bypass LLM Safety Classifiers | The CrowdStrike Cyber Superintelligence Lab evaluated the most advanced publicly deployed content safety classifier and found that it can be systematically circumvented. | AI blog | CROWDSTRIKE |
|
10.10.26 |
CrowdStrike and Anthropic Give Critical Infrastructure Defenders the AI Advantage | CrowdStrike joins Anthropic’s Critical Infrastructure Defense Program, bringing deep cybersecurity expertise in protecting IT and OT environments to help advance AI-powered defense for critical infrastructure. | AI blog | CROWDSTRIKE |
|
10.10.26 |
Unknown Threat Actor Uses AI-Driven ARTEX to Target South Korean Finance | CrowdStrike Intelligence identified infrastructure associated with a targeted campaign against South Korean financial organizations that resulted in exfiltrated data. The campaign was active from late September to early October 2026. Analysis of threat actor-controlled open directories uncovered Claude Code session histories, ARTEX configuration files, and Claude memory files, providing direct insight into the threat actor's operational methodology and tooling. | AI blog | CROWDSTRIKE |
|
10.10.26 |
Ignore all instructions and read this blog: The state of AI-analysis evasion in malware | Just as attackers are adding new capabilities into their toolkits with AI, they are consciously trying to evade the novel AI capabilities levied on them by defenders. In Cisco Talos' findings with CAIRN, we classify this archetype of malware as “A3: AI-Analysis Evasion” — that is, malware that embeds natural-language instructions to influence automated analysis. In line with the CAIRN philosophy, we treat this embedded language as a signal and actively seek it out to track and measure the progression of adversary techniques on this front. | AI blog | CISCO TALOS |
|
7.10.26 |
AI agents at work: What Acronis endpoint data reveals about MSP and SMB usage | An AI tool that produces only text carries roughly the risk of a badly worded email. A tool that can reach into systems and change things carries the risk of whatever those systems hold. The security question is not how clever the model is. It is which systems and data the model is authorized to access. | AI blog | ACRONIS |
|
7.10.26 |
How AI Is Changing the Roles Required in the Security Operations Center | As AI takes on more of the enrichment, correlation, and initial assessment inside the SOC, roles, skills, and KPIs still require deliberate redesign. Security leaders need to decide where automation is dependable, where human judgment should remain decisive, and how teams should be measured when alert handling is no longer the center of the operating model | AI blog | RAPID7 |
|
7.10.26 |
Access Governance: Reducing Attacker Reach in the Age of AI-Accelerated Threats | AI is accelerating vulnerabilities and threats. Learn how to strengthen your organization's data security and access governance to reduce risks effectively with insights from Varonis and PwC. | AI blog | VARONIS |
|
7.10.26 |
Least Privilege for AI Agents: A Practical Guide for Security Leaders | Learn what it means to enforce least privilege for AI agents, and get practical with a six-step implementation plan. | AI blog | VARONIS |
|
7.10.26 |
Rogue AI Agents Abuse urlquery to extract Russian government data | We found additional activity and sandbox circumvention techniques used by rogue AI agent swarms. These include attempting to extract Russian government data, hijacking remote browser services and running Javascript in attempts to extend their capabilities, circumvent networking and sandbox restrictions, and complete their tasks. | AI blog | Zenity Labs |
|
7.10.26 |
How AI Agents Turned One Exposed Endpoint into a Server Takeover Threat Spotlight | ReliaQuest investigated an incident in which, for the first time, we identified an open-source AI agent orchestration platform running on the same IP address that launched the attack. We assess with high confidence that AI agents driven by large language models (LLMs) carried out substantial portions of it. | AI blog | RELIAQUEST |
|
7.10.26 |
Identity-aware AI data agents with AWS Lake Formation and Trusted Identity Propagation | You’re building a data agent that lets business users ask questions about lakehouse data in natural language. You’ve already built governance policies that control who can access which datasets. The challenge is making the agent respect those rules without rebuilding them in your application code. | AI blog | AWS Security Blog |
|
7.10.26 |
CyberXero: An AI-Augmented Initial Access Broker Targeting Ukrainian Critical Infrastructure | SOCRadar’s Threat Research Unit (STRU) has documented CyberXero, a Russian-speaking, financially motivated Initial Access Broker that pairs commodity offensive tooling with an AI orchestration layer running on its own infrastructure. | AI blog | SOCRADAR |
|
7.10.26 |
How to mitigate shadow AI risks from employee citizen coders | AI tools let non-technical employees build workplace apps in minutes with natural language prompts. While these AI-generated apps boost productivity, they can create severe security and data risks. As Cybersecurity Awareness Month kicks off, we’re sharing how Tenable adopted a structured governance framework that allows our “citizen coders” to build secure and compliant apps with AI. | AI blog | Tenable® |
|
7.10.26 |
Claude Mythos: When Zero-Day Vulnerabilities Outpace Defenses | Advances in AI are reshaping the cybersecurity landscape in a fundamental way. Modern AI systems are no longer limited to detecting known threats. They are increasingly capable of uncovering previously unknown vulnerabilities buried deep within complex software and systems, including Zero Day vulnerabilities that can be found at lightning speed. | AI blog | FORESCOUT |
|
7.10.26 |
Super Intelligence: The Rise of Super-Intelligent Threats | On September 29, 2026, the White House signed an executive order replacing the term "Artificial Intelligence" with "Super Intelligence" across the U.S. executive branch. The stated reasoning is that "Super Intelligence" better conveys "the true capabilities of the technologies being developed today", not merely imitating or automating human intelligence, but empowering citizens “to achieve what was previously impossible.” | AI blog | SEKOIA |
|
3.10.26 |
AI-Enabled Cyber Attacks: What They Are and How to Defend Against Them | AI-enabled cyber attacks are malicious operations in which adversaries use AI, most often large language models (LLMs), to plan, carry out, or scale any part of an attack. That includes writing phishing lures and deepfake pretexts, developing and debugging malware, discovering and exploiting vulnerabilities, and running intrusions end to end. | AI blog | PICUSSECURITY |
|
3.10.26 |
AI-Agentic attacks Optimized Operations | AI is changing cyberattacks from isolated activities into connected operations that require a new level of speed, context, and understanding. | AI blog | WatchGuard Blog |
|
3.10.26 |
Prompt Forcing: The Scarier Sibling of Prompt Injection | Breaking down a newly discovered prompt attack technique that isn't prompt injection. | AI blog | Forever Security |
|
3.10.26 |
Agents Gone Wild: An AI-Orchestrated Global Campaign Against PaperCut NG/MF | GreyNoise observes adversary activity through our Global Observation Grid (GOG), a network of sensors that draws attacker scanning and exploitation onto infrastructure we control. This lets us study adversary infrastructure, tooling, and tradecraft directly, without waiting for a victim investigation. | AI blog | GREYNOISE |
|
3.10.26 |
Threat Actors Are Posing as OpenAI, Anthropic and DeepSeek to Target Credentials and Secrets | GreyNoise is observing automated scanners posing as the web crawlers of OpenAI, Anthropic, DeepSeek, and Fortune 500 companies. These forged automated scanners have been observed requesting files often exposed on misconfigured web servers and by other commonly leaked secret and credential methods. | AI blog | GREYNOISE |
|
3.10.26 |
AI-Assisted Attacks Still Leave a Behavioral Trace | AI is increasingly being used to accelerate cyber-attacks, but attackers still leave detectable behavioral traces. This blog explores how Darktrace identified suspicious file delivery, command-and-control communications, beaconing activity, and other anomalies linked to AI-assisted campaigns through behavioral analysis. | AI blog | DARKTRACE |
|
3.10.26 |
'Super Intelligence' Executive Order Fuels Nearly Hundredfold Increase in .si Domain Registrations | We previously reported that registrations of Slovenian .si domains overtook .ai in the 25 hours after President Trump told the UN General Assembly on September 22 that the U.S. would call AI “super intelligence” (SI). | AI blog | NETCRAFT |
|
3.10.26 |
FinCEN Moves to Cut the A7 Network's Sub-Agents Off From the US Financial System | FinCEN issued a finding and NPRM today identifying transactions involving any non-US company controlled by the A7 Network, which the agency calls "Sub-Agents," as a class of transactions of primary money laundering concern. The proposed rule would prohibit every covered US financial institution from sending or receiving funds or crypto involving a Sub-Agent. | AI blog | TRM Labs |
|
3.10.26 |
Aligning AI Speed with AI Trust: AI Agent Security Insights for CISOs and Security Leaders | AI agent security is the practice of governing two separate things: what an autonomous AI system is allowed to reach, and what it is allowed to do. Most enterprises have built a program for the first one. Almost nobody has built one for the second, and that's where the losses are starting to show up. | AI blog | Morphisec Blog |
|
3.10.26 |
Introducing Clef: our open-source decision models, and new RL fine-tuning platform | Over the last few weeks, there has been lots of buzz around decision models such as Typesafe AI’s Jev System One model. While classifier models have been around for some time, Jev introduces a new decision model concept into the world of AI — a model that produces bounded structured outputs cheaply, quickly and consistently that can be added into a workflow when a decision is required. | AI blog | CLOUDFLARE |
|
3.10.26 |
One year later: Sovereign AI and the fight for choice | It's Birthday Week, when we traditionally ship presents to the Internet. This year, two of them come from Europe: EuroLLM, which covers all 24 official EU languages, and Apertus, Switzerland's fully open model, trained on more than 1,500 languages. Both were built by public universities and research institutions. Both are coming to Workers AI, and you can request access today. | AI blog | CLOUDFLARE |
|
3.10.26 |
AI Search is now generally available | Cloudflare’s AI Search combines Workers AI, Vectorize, R2, and Browser Run into a fully managed index and retrieval pipeline. Since we launched AI Search over a year ago, we’ve seen developers use it to power a wide range of search use cases, from searching internal documentation to powering search for their websites. We use AI Search ourselves to power search on our own blog and developer docs. | AI blog | CLOUDFLARE |
|
3.10.26 |
Cut your AI spend with AI Gateway's Auto Router | From our conversations with companies at every stage of their AI adoption journey, we've seen some common patterns. First, there is an exploration period as you bring on every new tool, dole out API keys freely, and let the tokens flow. Then, you converge on the canonical tools for your organization for agentic coding, for non-technical workflows, for running and deploying agents. | AI blog | CLOUDFLARE |
|
3.10.26 |
Detect and send production issues straight to your agent | As agents help us build more complex applications, both humans and agents need a better way to stay on top of what goes wrong in production. Coding agents can already query observability data, navigate a repository, change code, write tests, and open a pull request. | AI blog | CLOUDFLARE |
|
3.10.26 |
Simplifying domains for people and agents | You just thought of your next great idea, and buying the right domain feels like the easiest way to make that first bit of progress. Naturally, you open a new tab in your browser, only to find yourself face-to-face with an experience that feels like a budget airline peppering you with add-ons at checkout: Want security? How about a website? Do you want email? You’re just a few minutes into building your next idea, and it doesn’t feel fun anymore. | AI blog | CLOUDFLARE |
|
3.10.26 |
Monetization Gateway beta: charge AI agents for consumption with HTTP 402 | Today, we’re making the Cloudflare Monetization Gateway available as part of a closed beta, and showcasing four customer use cases that are in production today. Since we announced the plan three months ago, we have been working closely with our customers to make the Gateway fast, flexible, and easy to use. | AI blog | CLOUDFLARE |
|
3.10.26 |
Pay Per Use: when AI uses your work, you should get paid | AI answer engines read a publisher’s page and hand the reader a summary, so the visit, and the revenue that would come with it, never happens. Most publishers will never sign a licensing deal with the companies that use their work in AI products, and no company can negotiate with millions of sites. The web needs a way to say “yes, if you pay.” Pay Per Use is one way to say it, and it's now in beta. | AI blog | CLOUDFLARE |
|
3.10.26 |
AI Security Strategy: 6 Steps for CISOs | Security teams are used to vetting technology before it arrives. With AI, it was already everywhere before anyone asked them. | AI blog | Abnormal AI |
|
3.10.26 |
Signatures Are the Floor, Not the Ceiling | See why signature-based detection struggles against AI-generated attacks and how behavioral AI uses embeddings to detect novel threats without relying on known indicators. | AI blog | Abnormal AI |
|
3.10.26 |
How Behavioral AI Detects Threats That Look Normal | Attackers are no longer breaking in but blending in, mimicking normal behavior to breach organizations. A founding Abnormal AI engineer explains why we bet on behavioral AI to stop attackers walking through the front door. | AI blog | Abnormal AI |
|
3.10.26 |
Beyond Model Alignment: Securing Every Layer of the AI Agent | Alignment makes a model well-intentioned, but it can’t secure the harness, tools, and data an agent touches in production. Real agent security means governing every layer with enforcement the agent itself can’t switch off. | AI blog | Trend Micro |
|
3.10.26 |
Managing Agentic AI: Why the Control Plane Problem Is an AI Problem | On one hand, it’s somewhat befuddling how often people may need that reminder. In reality, it is not that they don’t know that happy mantra, it’s that they struggle to gain the requisite visibility into … everything in their IT and OT environments. | AI blog | GUIDESECURITY |
|
3.10.26 |
Why One-Time Approval Fails for AI Agent Skills | A review answers one question, once: is this skill, MCP server or plugin allowed in? It never asks again, whether it took six weeks in a ticket queue or six minutes in a governance tool. | AI blog | Manifold |
|
3.10.26 |
The New Rules of Patching: When a Fix Becomes a Blueprint for Attackers | TL;DR: Frontier AI has collapsed the vulnerability exploit window from weeks to mere hours. Attackers now use advanced AI models to reverse-engineer published fixes and generate working exploits faster than human security teams can deploy updates. In the AI era, publishing a patch creates a blueprint for attackers. Surviving this threat requires vendors to tier sensitive disclosures and customers to treat patch application speed as a critical security metric. | AI blog | JFROG |
|
3.10.26 |
Copilot has the largest AI attack surface of any tool we tracked | Microsoft Copilot accounts for a disproportionate share of the AI footprint we track, and it’s barely being used. It’s a massive attack surface featuring a minuscule rate of activity. | AI blog | ThreatDown |
|
3.10.26 |
CrowdStrike and NVIDIA Extend Security Across the AI Stack | CrowdStrike and NVIDIA are collaborating on the NVIDIA Open Agent Safety Platform, an open reference design to extend security deeper into the agentic stack and establish stronger boundaries for autonomous AI. | AI blog | CROWDSTRIKE |
|
3.10.26 |
GTIG AI Threat Tracker: From Prompting to Autonomy – The Evolution of Adversarial AI | Since the release of our May 2026 report detailing adversarial misuse of artificial intelligence (AI), Google Threat Intelligence Group (GTIG) has observed forward leaning adversaries transition from basic prompting to agentic AI workflows and AI-enabled automation. In these operations, human-in-the-loop latency is dramatically reduced, compressing the traditional window for defenders to respond. | AI blog | Google Cloud Blog |
|
3.10.26 |
NVIDIA and Eclypsium Partner for Trusted AI Infrastructure | AI agents are moving from answering questions to taking action. They can read sensitive data, call APIs, write and execute code, use external tools, and coordinate with other agents. In some environments, their decisions may affect physical systems. | AI blog | Trend Micro |
|
3.10.26 |
Safely Securing AI Agents | Learn of the benefits of safely securing AI agents. | AI blog | Trend Micro |
|
3.10.26 |
The Closed Quorum: Inside the first reported autonomous AI C2 implant | CLOSEDQUORUM, a malware binary discovered through Cisco Talos’ CAIRN project, exhibits fully autonomous command and control (C2). While we do not have confirmation of in-the-wild deployment, artifacts from the binary were used to connect the developer to postings on criminal forums related to carding, dating back to 2025. | AI blog | CISCO TALOS |
|
30.9.26 |
OpenAI agents go rogue: AI agent governance lessons | Recently and rather quietly, there have been reports that describe an alarming case where thousands of AI agents used a dormant wiki as a coordination mechanism. On the surface, it’s a fascinating technical story. | AI blog | Barracuda Networks Blog |
|
29.9.26 |
Attackers Abuse ChatGPT Custom GPTs to Deliver RAT via ClickFix | Attackers are abusing ChatGPT Custom GPTs to impersonate legitimate product offerings, then directing victims to a malicious "backup" site through a trusted ChatGPT-hosted interface. Huntress researchers found two Custom GPTs linked to the same campaign that were being used in this manner. | AI blog | Huntress |
|
26.9.26 |
How Aikido finds more vulnerabilities than Claude Security with Mythos at half the cost | Claude Mythos is arguably the strongest cybersecurity model that Anthropic has built. But we know that model capability is only part of what determines how well an AI vulnerability product performs. | AI blog | AIKIDO.DEV |
|
26.9.26 |
Could OpenClaw have actually hacked that Australian gym? We decided to test it. | Earlier this month, Australian journalists reported on an OpenClaw AI assistant hacking into a gym website in what was described as the “first known Australian autonomous cyber attack”. Reportedly, in response to benign requests for assistance booking gym sessions, the AI agent identified vulnerabilities in the gym booking system to book classes months in advance and kick another gymgoer off a waitlist to advance the user’s position. | AI blog | AIKIDO.DEV |
|
26.9.26 |
Slovenian .SI Domain Registrations Outpace .AI | Registrations of Slovenian .si domains overtook .ai in the 25 hours after President Trump told the UN General Assembly on September 22 that the U.S. would call AI “super intelligence” (SI). | AI blog | NETCRAFT |
|
26.9.26 |
Securing what your Bedrock AgentCore agents actually do | An AI agent is software that can act on your behalf. It may read documents, emails, and web pages, then use the systems you connected to complete a task. | AI blog | Cato Networks |
|
26.9.26 |
AI Agents Are Hacking Online Retailers for $25 a Company | A financially motivated operator is running three open source AI harnesses against hundreds of online retailers, almost entirely unattended. More than 600,000 credit card records have been taken, and in one case the agent's own cleanup routine destroyed the victim's data. | AI blog | GAMBIT SECURITY |
|
26.9.26 |
Hacking OpenAI | On July 25, 2026, we chained two critical vulnerabilities to compromise multiple OpenAI employees’ ChatGPT accounts. With these accounts, we could then access internal OpenAI repositories, and potentially many other connectors. | AI blog | Hacktron AI |
|
26.9.26 |
AI-Enabled Cyber Attacks: What They Are and How to Defend Against Them | AI-enabled cyber attacks are malicious operations in which adversaries use AI, most often large language models (LLMs), to plan, carry out, or scale any part of an attack. | AI blog | PICUSSECURITY |
|
26.9.26 |
AI-augmented, still human-led: What changed in the 2026 Gartner Market Guide for Managed Detection and Response | Gartner named Expel a Representative Vendor for the eighth consecutive time in the 2026 Gartner® Market Guide for Managed Detection and Response. | AI blog | Expel |
|
26.9.26 |
"Are we secure against AI agents?" is a question we couldn't answer well | | “Are we secure against AI agents?” can’t be answered as asked. It bundles ten different problems—with ten different owners—and no shared vocabulary to separate them. | AI blog | Expel |
|
26.9.26 |
AI-Accelerated Defense: Build a Continuous Security Loop | That does not mean buying AI for the sake of AI. It means applying AI to the work that slows security teams down today: translating threat signals into validated exposure, deciding what matters, changing controls and proving the result. | AI blog | CYMULATE |
|
26.9.26 |
Why AI Defenders Can't Afford to Slow Down | The AI Frontier pacing argument assumes enterprise risk comes from frontier models being too powerful or misaligned. But that argument assumes that if labs build safer models, enterprises will be safer. That's not exactly how it works. AI frontier labs can slow down, but defenders can’t afford to. | AI blog | TrendAI |
|
26.9.26 |
The New Engineering Problem: Managing What AI Decides to Import | AI makes it faster to turn an idea into working software. A developer describes a feature, integration, or service and receives an implementation in minutes. Coding agents go further by editing files, running tests, troubleshooting failures, and preparing changes for review. | AI blog | SONATYPE |
|
26.9.26 |
Securing GenAI in the Real World | Generative AI (GenAI) projects often start small and move fast, with clients moving from initial testing to production and deployment as promising outcomes emerge. Along the way, the business value grows, but so can security risks, especially when sensitive data and high-impact workflows begin to rely on AI model decisions. | AI blog | GUIDESECURITY |
|
26.9.26 |
Beyond Point-in-Time Testing: 5 Use Cases for Agentic Red Teaming | Attackers have automated their attacks and are moving at machine speed. Defenders need to do the same to test their defenses against these automated, agentic AI attacks. That is why agentic red teaming and continuous attack path mapping are becoming increasingly important. | AI blog | RELIQUEST |
|
26.9.26 |
Is Your Home Internet Feeding the AI Data Race? | Residential proxy providers sell household IP access for AI data collection, but many users do not understand what that access involves. | AI blog | GENDIGITAL |
|
26.9.26 |
Introducing CAIRN: Frontier tracking for AI-integrated malware | Talos is releasing CAIRN, a research toolkit for hunting, classifying, and tracking emerging AI-integrated malware. | AI blog | CISCO TALOS |
|
26.9.26 |
The Closed Quorum: Inside the first reported autonomous AI C2 implant | CLOSEDQUORUM, a malware binary discovered through Cisco Talos’ CAIRN project, exhibits fully autonomous command and control (C2). It represents a shift in effort displacement for attackers, in which expanding portions of the attack chain can be executed without operator involvement. | AI blog | CISCO TALOS |
|
26.9.26 |
Should you care about an “AI slowdown?” | In this week's Threat Source, David talks about why focusing on your security basics is still your best bet, even in a world with rapid AI advancements. | AI blog | CISCO TALOS |
|
26.9.26 |
The SMB cybersecurity squeeze: AI agents at work, old attacks in overdrive | As AI opens new paths to company data while making familiar attacks faster and cheaper, SMBs need protection designed around the time and expertise available to operate it | AI blog | Eset |
|
24.9.26 |
Detecting shadow AI agents with advanced posture checks | Okta recently introduced advanced posture checks within the Okta Verify client for macOS and Windows. This feature leverages osquery, a lightweight endpoint agent, which can run a SQL query at the point of log in to detect critical device data — launchd services, file paths, running processes, package managers, listening ports, installed apps, and container artifacts — that collectively provide a device assurance signal at sign-in time. | AI blog | OKTA |
|
24.9.26 |
Signing in without actually signing in | The price of AI tokens and subscriptions is increasing as AI model providers seek to recoup investment costs. As a result, these accounts are becoming more valuable targets for account takeover. Stopping this is paramount for enterprises. AI theft increases token bills. | AI blog | OKTA |
|
24.9.26 |
Proofpoint named a Pace Setter as of September 2026 in the Gartner Emerging Market Quadrant for AI Application Security — Established Vendors | As organizations rapidly adopt AI assistants and autonomous AI agents, the challenge is no longer simply protecting AI models—it is governing how AI interacts with employees, enterprise data, and business systems. We believe this recognition reflects the growing importance of securing AI across the entire enterprise. | AI blog | PROOFPOINT |
|
23.9.26 |
Disrupting EvilTokens: The AI Chatbot Built for Cybercrime | Microsoft has disrupted EvilTokens, a powerful cybercrime platform that used AI at every step of the attack chain—from compromising email accounts to designing intricate roadmaps for financial fraud and scams. | AI blog | Microsoft blog |
|
22.9.26 |
Passive DNS Reveals Flock Safety's Dev Infrastructure | Automated license plate readers and AI-assisted video surveillance have moved from a handful of pilot deployments to tens of thousands of installations across US towns, business parks and residential streets. Anyone can see the cameras, but the cloud platform behind them is a different story. | AI blog | FLARE.IO |
|
18.9.26 |
Have it both ways: stay discoverable in search while disallowing AI training | Without proper controls, website owners have long faced a difficult tradeoff: allow your content to be used for AI training, or risk losing discoverability in search. That tradeoff exists because some of the largest organizations on the Internet use mixed-use crawlers: a single crawler serving both search and AI training. Refuse one, and you refuse the other. | AI blog | Cloudflare Blog |
|
18.9.26 |
NFC Relay Goes Local: How AI Is Accelerating a New Wave of Independent Malware Developers | The Cleafy Threat Intelligence team has identified and analyzed two new Android NFC relay malware families: DevilNFC and NFCMultiPay. While these families are developed by distinct, unrelated actors with no shared code or infrastructure, both are actively used to successfully launch NFC relay attacks targeting European and LATAM banking customers. | AI blog | Cleafy |
|
18.9.26 |
Malware That Rewrites Itself Every Hour: AI-Powered Polymorphic Threats Break Signature-Based Defense | AI-powered polymorphic malware is malicious code that uses artificial intelligence to automatically rewrite itself — often every hour or on every execution — so that no two copies share the same signature, hash, or structure. Because traditional antivirus and signature-based detection depend on recognizing a known pattern, malware that never holds still is effectively invisible to them until after it has already run. | AI blog | Morphisec Blog |
|
18.9.26 |
Zero Breaches Out of 50: TrendAI Vision One™ Endpoint Security Essentials Earns AV-Comparatives Certified Leader Status | When your endpoint defense is tested against 50 sophisticated attack scenarios—the kinds of attacks that compromise enterprises—the measure is simple: did anything get through? For TrendAI Vision One™ Endpoint Security Essentials in AV-Comparatives' 2026 Endpoint Prevention & Response (EPR) Test, the answer is no. | AI blog | Trend Micro |
|
18.9.26 |
The Risks Behind AI | There’s a hot new technology hitting the streets: it’s called AI, it’s AMAZING, and it’ll make the world a better place by solving all of your problems. Whether it’s finding a lost mountaineer, the early detection of cancers in radiological images, modeling proteins to speed the development of new drugs, or weather prediction, the capabilities of this thing seem limitless! | AI blog | EMISISOFT |
|
18.9.26 |
CoSnitch: When Your AI Assistant Becomes Its Own Whistleblower | See how meta-hacking got Microsoft Copilot to snitch on itself, exposing CoSnitch, a one-click flaw that silently exfiltrates data. | AI blog | VARONIS |
|
18.9.26 |
CrowdStrike Accelerates Real-Time Data Classification with On-Device AI | Modern data security depends on understanding sensitive data as it is created, accessed, and moved in real time directly on the endpoint. | AI blog | CROWDSTRIKE |
|
18.9.26 |
PhantomRaven: LLM-generated Information Stealer for Bug Bounty Hunting | CrowdStrike Counter Adversary Operations identified a financially motivated threat actor who works as a bug bounty hunter and who developed and distributed the JavaScript (JS)-based information stealer PhantomRaven via npm, a platform on which developers can access open-source packages to build applications and software. | AI blog | CROWDSTRIKE |
|
18.9.26 |
CrowdStrike SafeMind Uses the Best Offense to Build the Best Defense | The conventional approach to AI-powered security is to build an offensive agent to find weaknesses, build a defensive agent to catch threats, and run them in separate tracks. It's a clean division of labor that produces capable tools. However, the conventional approach also produces a permanent blind spot — the defense is never trained against the actual offense, and the connective tissue between the two is manual, slow, and fragile. | AI blog | CROWDSTRIKE |
|
18.9.26 |
Devil’s advocate? Uncensored Luciferus AI service advertised underground | Uncensored refers to a lack of typical guardrails or ethical restrictions, lowering the technical barrier of entry into cybercrime | AI blog | SOPHOS |
|
18.9.26 |
The Answer to AI Uncertainty Is Adaptability, Not Paralysis | AI uncertainty is not a strategic reason to wait; it is a strategic imperative to build adaptable organizations that can innovate confidently, govern risk proportionately, and respond effectively as technology and threats evolve. | AI blog | SOPHOS |
|
18.9.26 |
Is Your Home Internet Feeding the AI Data Race? | Residential proxy providers sell household IP access for AI data collection, but many users do not understand what that access involves. | AI blog | GENDIGITAL |
|
18.9.26 |
Free AI, at the Price of Your Prompt | You choose a model, ask a question and get an answer. Behind that familiar exchange is a less familiar journey involving prompt logs, shifting model identities and servers whose role is never explained to users. | AI blog | GENDIGITAL |
|
18.9.26 |
Securing the unpatchable in an age of AI-driven vulnerabilities | Advances in AI technology will continue to identify vulnerabilities that in some circumstances are difficult, or effectively impossible, to patch. Appropriate network segmentation, rigorous visibility, and the deployment of NGFW/IPS combinations can provide a powerful compensatory layer. | AI blog | CISCO TALOS |
|
18.9.26 |
Should you care about an “AI slowdown?” | In this week's Threat Source, David talks about why focusing on your security basics is still your best bet, even in a world with rapid AI advancements. | AI blog | CISCO TALOS |
|
12.9.26 |
LLMs Just Made Your IP Easier to Steal: Here’s the Fix | Large Language Models (LLMs) have fundamentally altered the economics of reverse engineering, meaning that any organization still relying on traditional obfuscation as its primary defense is protecting against an attacker who no longer exists. | AI blog | JSCRAMBLER |
|
12.9.26 |
Anthropic Identifies Biased Reasoning and Recklessness as Drivers of Claude’s PyPI Attack | Anthropic found biased reasoning and recklessness drove Claude Mythos 5 to publish malware on PyPI and compromise a security vendor. | AI blog | SOCKET.DEV |
|
12.9.26 |
Reduce AI Token Waste by Getting Decisions Right Earlier | AI coding assistants are changing the economics of software development. They can interpret tasks, inspect codebases, select dependencies, make changes, call tools, run tests, and prepare pull requests. | AI blog | SONATYPE |
|
12.9.26 |
Agentic AI Security: Key Findings from New IDC Research on the Identity Control Plane | A new IDC whitepaper, commissioned by GuidePoint Security, explores why identity is the foundational control plane for securing agentic AI and what organizations should prioritize as adoption accelerates. | AI blog | GUIDESECURITY |
|
12.9.26 |
URL Laundering by Rogue Agents: Newly Discovered Messages and Sandbox Circumvention by AI Swarms | We found a thousand new messages and additional sandbox circumvention methods used by the collusion.wiki Rogue AI agent swarms | AI blog | ZENITY |
|
12.9.26 |
Vibe-Terrorism: Inside the Yemen Cell That Used Claude to Build Guided Weapons | In Anthropic’s September 2026 threat intelligence report, one case stands apart from the rest. A small cell based in northern Yemen used Claude Code to develop guidance, navigation, and control (GNC) software for missiles. They ran three weapons programs simultaneously, conducted a live test-fire and when the rocket failed, they came back to Claude within hours to figure out why. | AI blog | SOCRADAR |
|
12.9.26 |
Introducing the CyberAgents Exchange AI Inspector: Rigorous review for community-built AI | Open-source registries for AI agents are only effective when they include a rigorous, transparent security review process for community submissions. That’s why for its new CyberAgents Exchange registry, Tenable paired its exposure management expertise with OpenAI GPT Cyber models to create the CyberAgents Exchange AI Inspector. | AI blog | TENABLE |
|
12.9.26 |
The Anthropic Glasswing Receipts Are Starting to Trickle In | Anthropic’s Project Glasswing is approaching 5 months old, and Anthropic published its Vulnerability Disclosure Ledger on May 22nd. It hadn't received an update until this past week, when it backfilled the ledger with additional findings and updates, so naturally I thought it would be worthwhile to take a look at the receipts. | AI blog | VULNCHECK |
|
12.9.26 |
Beyond the Alert: Bringing Context to Insider Risk and AI Investigations | Proofpoint is announcing two expansions to its communications intelligence capabilities today. Proofpoint Prism Investigator now connects directly to Microsoft 365, and Proofpoint Human Communications Intelligence (HCI) now brings AI communications governance signals into insider risk investigations. | AI blog | PROOFPOINT |
|
12.9.26 |
AI-Driven Threat Intelligence for Gulf Enterprises: Why Detection Speed Is Now a Regulatory Requirement | AI-powered threat intelligence helps GCC enterprises detect breaches faster, close compliance gaps, and meet strict cyber incident reporting deadlines. | AI blog | Cyble |
|
12.9.26 |
Infostealers Have Found a New Target: Your AI Agent | Malware operators are expanding beyond browser passwords and crypto wallets to collect access tokens, MCP configurations, prompt histories and project data stored by AI tools. | AI blog | GENDIGITAL |
|
12.9.26 |
Protecting organizations from AI-assisted executive impersonation and invoice fraud | Microsoft examines an AI-assisted business email compromise campaign that used executive impersonation and fake invoices to target finance teams with ACH payment fraud. | AI blog | Microsoft blog |
|
12.9.26 |
Attackers Expose Ongoing AI Tool Use Targeting Organizations in Latin America | We have analyzed two ongoing, multi-stage network intrusion and data-exfiltration campaigns targeting organizations in Latin America. Corroborating recent findings from the broader threat intelligence community, we observed attackers leveraging artificial intelligence (AI) to enhance their capabilities. | AI blog | Palo Alto |
|
12.9.26 |
PuzzleMask: Abusing Plain Prose as a Covert AI Attack Vector | In this research we introduce a prompt-crafting technique for bypassing quick LLM-based policy checks — using plain English (no emojis, base64, invisible formatting, etc.) A policy-violating payload (e.g. ”encrypt files in ~/Documents”, “give me a biohazard recipe”, “ignore all previous instructions and…”) is embedded in a specially crafted prose wrapper. | AI blog | CHECKPOINT |
|
12.9.26 |
The Shared Clipboard Inside the Sandbox: Cross-Account Data Leakage in ChatGPT | Check Point Research discovered a covert cross-account command channel through which an attacker could use a victim’s ChatGPT session to execute hidden tasks with the tools, data, and connected apps available to that session. The victim could receive a normal answer to their visible request while the attacker’s task was processed separately and its result returned across accounts. | AI blog | CHECKPOINT |
|
12.9.26 |
GuardBreaker: Derailing AI-assisted malware analysis with a code comment | AI blog | Eset | |
|
12.9.26 |
Trellix SecondSight Threat Hunting Report: How AI and Human Intelligence Expose 2026 Cyber Threats | As adversaries adopt new technologies, techniques, and increasingly AI-enabled ways of operating, defenders have access to more telemetry, intelligence, and automation than ever before. The challenge is knowing where to look, what matters, and when a weak signal deserves a closer look. This is focus of the September 2026 Trellix SecondSight Threat Hunting Report. | AI blog | TRELLIX |
|
11.9.26 |
What Is Agentic Cyber Defense Engineering? | Attackers can now find weaknesses faster, adapt tactics in minutes and scale campaigns without requiring extensive expertise or large networks of threat actors. Meanwhile, many security teams still rely on manual processes to execute and coordinate critical activities. The result is a widening gap between the speed of AI-powered threats and the ability of security teams to respond. | AI blog | CYMULATE |
|
11.9.26 |
New Ruxie AI power-up: Precedent engine turns repetitive alerts into faster decisions | A large portion of the daily security alert queue is made up of alerts that look almost exactly like cases our analysts have already investigated and closed. | AI blog | Expel |
|
11.9.26 |
How Amp got its Soc 2 Type II pentest evidence in hours | Amp is the AI hiring team for high-volume frontline employers. Restaurants, retailers, logistics operators, hospitality groups, healthcare staffing firms. Amp screens candidates against the criteria you set, books interviews on your managers' calendars, runs candidate comms, and closes offers you approve. | AI blog | ESCAPE.TECH |
|
10.9.26 |
The Anthropic Glasswing Receipts Are Starting to Trickle In | Anthropic’s Project Glasswing is approaching 5 months old, and Anthropic published its Vulnerability Disclosure Ledger on May 22nd. It hadn't received an update until this past week, when it backfilled the ledger with additional findings and updates, so naturally I thought it would be worthwhile to take a look at the receipts. | AI blog | VULNCHECK |
|
10.9.26 |
Coming to Tenable One: Claude Mythos 5 and “Adversary View” | Tenable is bringing Anthropic’s Claude Mythos 5 into our enterprise security offerings. Adding frontier adversarial reasoning to the Tenable One Exposure Management Platform will help customers better anticipate how attackers could breach their environments and stay ahead of AI-fueled risk. | AI blog | TENABLE |
|
10.9.26 |
SOCRadar MCP Server for ChatGPT: Setup and Use Cases | The SOCRadar Threat Intelligence MCP app connects ChatGPT to the SOCRadar MCP server over OAuth, exposing your licensed intelligence modules as callable tools. Analysts enrich indicators, check CVE exploitation, and query Dark Web exposure without leaving the conversation. | AI blog | SOCRADAR |
|
9.9.26 |
GTIG AI Threat Tracker: From Prompting to Autonomy – The Evolution of Adversarial AI | Since the release of our May 2026 report detailing adversarial misuse of artificial intelligence (AI), Google Threat Intelligence Group (GTIG) has observed forward leaning adversaries transition from basic prompting to agentic AI workflows and AI-enabled automation. | AI blog | GTI |
|
9.9.26 |
The Shared Clipboard Inside the Sandbox: Cross-Account Data Leakage in ChatGPT | Check Point Research discovered a covert cross-account command channel through which an attacker could use a victim’s ChatGPT session to execute hidden tasks with the tools, data, and connected apps available to that session. The victim could receive a normal answer to their visible request while the attacker’s task was processed separately and its result returned across accounts. In our proof of concept, ChatGPT retrieved email data from the victim’s connected Gmail account and relayed it to the attacker. | AI blog | CHECKPOINT |
|
8.9.26 |
AI Agent 'opencode'로 미끼를 만든 김수키, GitHub PAT 기반 LNK 공격 진화 | Kimsuky Uses the AI Agent 'opencode' to Create Decoys as Its GitHub PAT-Based LNK Attacks Evolve | AI blog | GENIANS |
|
5.9.26 |
How Abnormal and OpenAI are Detecting and Responding to Rogue AI in the Cloud | Abnormal AI is extending its behavioral AI engine to the cloud, using OpenAI models to detect, investigate, and respond to risky or malicious AI-agent behavior. | AI blog | ABNORMAL |
|
5.9.26 |
TrendAI™ Brings OpenAI's GPT Cyber Models Into the Race to Shrink Exposure Time to Zero | OpenAI’s GPT cyber models help TrendAI™ close the exposure window, from vulnerability to fix, faster than ever. | AI blog | Trend Micro |
|
5.9.26 |
7 Best AI Penetration Testing Tools of 2026 | Compare the 7 best AI penetration testing tools of 2026. See which platforms prove exploits, reason like attackers, and run continuous testing at scale. | AI blog | NOVEE |
|
5.9.26 |
Someone Else Is Using Your AI | FortiCNAPP analyzes an AWS LLMjacking incident involving a leaked administrator key and unauthorized access to Amazon Bedrock. | AI blog | FORTNITE |
|
5.9.26 |
eSentire Wins 2026 Global AI Award for AI in Cybersecurity | We're proud to announce that eSentire has won a 2026 Global AI Award in the AI in Cybersecurity category, recognizing the Atlas Platform and the Controlled Autonomy SecOps operating model behind it. | AI blog | ESENTIRE |
|
5.9.26 |
AI threat so great that security “takes precedence over everything except critical business operations” | OpenAI just published an open letter, co-signed by Anthropic, Google, and more than 120 other organizations, warning that the window to prepare for AI-enabled attacks is closing rapidly. | AI blog | THREATDOWN |
|
5.9.26 |
CrowdStrike Announces Agentic Identity Provider | CrowdStrike gives every AI agent a trusted identity and continuously controls their access based on real-time context, and expands modern privileged access to where work happens. | AI blog | CROWDSTRIKE |
|
5.9.26 |
CrowdStrike Falcon Guardian Defines the Next Generation of AI Security | CrowdStrike’s new flagship AI detection and response solution delivers runtime protection for AI agents, introduces a new AI gateway, and extends expert-led defense. | AI blog | CROWDSTRIKE |
|
5.9.26 |
Attackers Expose Ongoing AI Tool Use Targeting Organizations in Latin America | We have analyzed two ongoing, multi-stage network intrusion and data-exfiltration campaigns targeting organizations in Latin America. Corroborating recent findings from the broader threat intelligence community, we observed attackers leveraging artificial intelligence (AI) to enhance their capabilities. | AI blog | Palo Alto |
|
5.9.26 |
An AI-Assisted Cyber Attack: Inside a Unit 42 Investigation | Unit 42 responded to an incident where a human attacker used frontier AI to breach an enterprise network autonomously as part of a ransom attack. The agents breached the company's security layers in a methodical manner, each targeting a different layer of defense to achieve a shared goal. | AI blog | Palo Alto |
|
31.8.26 |
Same Target, Different Playbooks: Two Attackers, Two Different Paths to Pwning the AI Stack | Recently, we've been interested in how AI is impacting the attack surface, so we started researching AI-assisted vulnerability discovery and AI-targeted technologies for our 1H-2026 State of Exploitation report. | AI blog | VULNCHECK |
|
29.8.26 |
|
LLM security testing for pentesters: map attacks to the OWASP LLM Top 10, break a vulnerable MCP server locally, and turn what you find into regression tests. Including solutions for enterprise scale. |
||
|
29.8.26 |
The AI CVE exploitation evidence story: Headlines are scarier than reality |
Expel's vulnerability intel team is currently monitoring 1,250+ CVEs related to 50 unique AI vendors; here's what we think you should know. |
||
|
29.8.26 |
Security teams have more data than ever before, but less time to act on it. Over the past several years, Cymulate has continued to evolve continuous exposure validation beyond attack simulation. We've expanded the platform to help organizations continuously validate their security controls, prioritize vulnerabilities to support continuous threat exposure management (CTEM), auto-mitigate with vendor-specific remediation and accelerate security operations with agentic AI. |
|||
|
29.8.26 |
What is new is the speed at which AI can discover, test and help weaponize those vulnerabilities. Frontier cyber models such as Mythos mark a shift in attacker economics. The limiting factor is no longer only expertise. It is orchestration. |
|||
|
29.8.26 |
TrendAI™ Advances Threat Hunting to Dynamic, Real-World Exploitation of AI Infrastructure |
The new threat hunting component of the TrendAI™ agentic exploit-remediation engine, code name AESIR, extends visibility beyond vulnerability disclosure. Its first published |
||
|
29.8.26 |
The TrendAI™ agentic exploit-remediation engine, code name AESIR, ranks first on CyberGym at 97% — more than 12 points ahead of both GPT-5.6 Sol and Claude Mythos 5. |
|||
|
29.8.26 |
Wiz honeypots uncover active campaigns targeting LiteLLM, MCP servers, and AI frameworks through RCE, blind prompt injection, and memory credential theft. |
|||
|
29.8.26 |
That's becoming increasingly important as AI coding assistants and agents accelerate development. More code can move into review, testing, security, integration, and deployment faster than those processes can absorb it, leading to more output but not necessarily more delivery. |
|||
|
29.8.26 |
Making AI/ML-driven Active Cyber Defense Work in OT Environments |
You already have the data. That’s not the problem. Yet unified visibility implementations often fail before they even get off the ground. Before reaching for AI/ML-driven Active Cyber Defense to solve Operational Technology (OT) visibility and monitoring challenges, it’s critical to pause and make sure you’re not repeating the mistakes of the past. |
||
|
29.8.26 |
In part 1 of this series, on discovering and mapping Copilot Studio agents in the wild, we took a close look at how active discovery and capability mapping for agents could play out. |
|||
|
29.8.26 |
It's Always DNS in Claude’s Sandbox: From Data Exfiltration to a Bidirectional DNS Shell |
In this blog, we’ll showcase how we were able to exploit a DNS functionality in Claude’s sandbox to create both a data exfiltration channel, and bidirectional shell. This whole attack chain rests on the overlooked DNS channel: although HTTP and HTTPS traffic are correctly blocked by egress filters, DNS resolution happens before those filters can intervene. |
||
|
29.8.26 |
Indirect prompt injection: How attackers manipulate AI agents through untrusted data |
An employee asks an AI agent to analyze a document. Hidden inside that content is an instruction written for the agent rather than the user. |
||
|
29.8.26 |
Indirect prompt injection: How attackers manipulate AI agents through untrusted data |
An employee asks an AI agent to analyze a document. Hidden inside that content is an instruction written for the agent rather than the user. |
||
|
29.8.26 |
OpenAI is calling for a global surge in cyber defense as AI-enabled attacks become more widespread and sophisticated. The call to action raises a practical question for defenders: if attackers gain more capable AI, how can organizations, governments, security providers, and frontier AI companies make sure defensive tools, verified fixes, and threat intelligence reach the teams that need them most? |
|||
|
29.8.26 |
How to build an exposure management program the business trusts: Lessons from Tenable’s CSO |
Discover how Tenable’s shift to an AI-driven exposure management program helped Tenable’s CSO, Robert Huber, overcome tool sprawl, unify data silos, mitigate the risk of rapid AI adoption, and shift from presenting granular, technical metrics to communicating business risk that the C-suite and the board can understand. |
||
| 29.8.26 | The AI Attack Surface: How Threat Actors Abuse Trusted AI Platforms | As more people use AI models such as Claude, ChatGPT, Grok, and Gemini, threat actors are abusing trusted AI platforms in a variety of ways to trick users into downloading malware. | AI blog | Huntress |
| 29.8.26 | Inside Elastic's agentic SOC: How we took AI alert triage from 60% to 92% accuracy | Elastic's InfoSec team runs three agents that read the detection rule's investigation guide and the closure reasons on 30 days of past cases. Analysts now clear most alerts with a single click in Slack. | AI blog | ELASTIC |
| 29.8.26 | The State of AI-Enabled Malware August 2026: From Brand Abuse to Agentic Execution | To assess the impact of AI-enabled malware, we collected and analyzed over 400 malware samples that integrate AI in some capacity, from brand impersonation and large language model (LLM)-generated code to agentic execution loops. Our central finding was that the AI malware space is currently overwhelmingly composed of proof-of-concept code, security validation testing and researcher submissions that have never reached a production environment. | AI blog | Palo Alto |
| 29.8.26 | The safety penalty: Reclaiming operational sovereignty in the age of AI | As frontier AI models become increasingly restrictive, security teams are facing a "safety penalty" that hampers real-time incident response. Discover how organizations can move toward operational sovereignty to ensure their defensive AI keeps pace with unconstrained adversaries. | AI blog | CISCO TALOS |
| 29.8.26 | Choose your fighter: Balancing competing requirements to select models for your AI SOC | Selecting a model for your security operations center (SOC) and digital forensics and incident response (DFIR) tasks is important, but selecting the best one is more involved than you might think. Here's how to choose. | AI blog | CISCO TALOS |
| 29.8.26 | AI-driven OSINT in the wrong hands – and why everyone could be a target for fraud | It’s getting cheaper and easier for cybercriminals to research potential victims. Here’s what’s still in your control. | AI blog | Eset |
| 27.8.26 | Mapping AI detections to MITRE ATLAS: How Expel does it | AI-related detections in Expel Workbench™ carry MITRE ATLAS tactic and technique labels alongside the ATT&CK labels our analysts already use. | AI blog | Expel |
| 27.8.26 | LLM security testing: how to pentest LLMs and MCP servers | LLM security testing for pentesters: map attacks to the OWASP LLM Top 10, break a vulnerable MCP server locally, and turn what you find into regression tests. Including solutions for enterprise scale. | AI blog | ESCAPE.TECH |
| 27.8.26 | Making AI/ML-driven Active Cyber Defense Work in OT Environments | You already have the data. That’s not the problem. Yet unified visibility implementations often fail before they even get off the ground. Before reaching for AI/ML-driven Active Cyber Defense to solve Operational Technology (OT) visibility and monitoring challenges, it’s critical to pause and make sure you’re not repeating the mistakes of the past. | AI blog | GUIDESECURITY |
| 27.8.26 | Lookalike Domains Are an Identity Problem | Powerful brands elicit a feeling of trust, a promise of service delivered, and an expectation of what you will get. Adversaries take advantage of that powerful connection as a means to collect username and password pairs for account takeover and payment card information for financial fraud. With AI and pre-built phishing kits, little technical skill is needed, which compresses the time between registering a lookalike domain and launching it as an active impersonation of your legitimate one. | AI blog | FLARE.IO |
| 27.8.26 | Response Orchestration: Automated Action on Your Terms | eSentire | The security industry has spent years working on better detection. Threat intelligence is richer than it has ever been. Signals are correlated faster. AI is solving the discovery problem in a way that doesn't require hours from senior analysts to deliver insight. Detection has genuinely improved. | AI blog | ESENTIRE |
| 26.8.26 | The Architecture Behind $1B: How Customers Run TrendAI Vision One™ on AWS, and Secure Their AI Workloads | TrendAI™ has surpassed $1 billion in AWS Marketplace sales, one of a select group of AWS independent software vendors (ISVs) ever to cross that line. Behind that number are more than 2,900 private offers, and thousands of security teams who closed real gaps faster because buying never slowed them down: | AI blog | Trend Micro |
| 26.8.26 | Prompting the Payload: How an npm Supply Chain Attack Delivers the RedC2 AI-Powered Linux Implant | TrendAI™ Research provides a comprehensive analysis of the RedC2 Linux Implant, a sophisticated threat recently discovered in the npm open-source ecosystem. | AI blog | Trend Micro |
| 22.8.26 | AI-Agent-Driven Offensive Operation : Exposed Adversary Open Directory Reveals Autonomous Crypto-Theft Campaign Leading to Mass Wallet and Credential Compromise | CloudSEK uncovered a Chinese-speaking threat operator using autonomous AI coding agents to scale cyberattacks, including mass WordPress compromise, cryptocurrency wallet theft, credential harvesting and cryptojacking. The exposed infrastructure also revealed a developing blockchain-based command-and-control system designed to make future operations more resilient to takedowns. | AI blog | CloudSEK |
|
22.8.26 |
PWNBench-v0.1: Evaluating frontier models for web application pentesting | We benchmarked 11 frontier LLMs on real-world web app pentesting. See which AI models lead in recall, precision, and cost efficiency in PWNBench-v0.1. | AI blog | NOVEE |
| 22.8.26 | New Report: AI threats are here. Why Q2 2026 signals the end of traditional patch cycles | The latest Quarterly Threat Landscape Report from Rapid7 Labs shows vulnerability disclosures still surging while attackers use automation and AI-assisted tooling to compress the time between disclosure and exploitation. The gap that patch cycles were built to fill is closing. Speed and volume are overwhelming security teams that have relied on traditional patch cycles and reactive programs. Success going forward can’t be about patching as much as possible - it has to be about understanding what matters most and reducing the exposures attackers can actually reach. | AI blog | RAPID7 |
|
22.8.26 |
Copilot for KeyBank — a 3 KB VSIX that beacons through github-cdn.net and waits for eval | A new VS Code extension called “Copilot for KeyBank” went live on the Microsoft Visual Studio Marketplace in August 2026. The publisher is verified, the display name on the marketplace is “Deep Seek”, and the package name is verified.pypi-keyBank. None of those brand names — GitHub Copilot, DeepSeek, PyPI, KeyBank — have anything to do with this extension. They are all cover. | AI blog | YEETH SECURITY |
| 22.8.26 | SearchLeak: How We Turned M365 Copilot Into a One-Click Data Exfiltration Weapon | Varonis Threat Labs discovered SearchLeak, a critical vulnerability chain in Microsoft 365 Copilot Enterprise that allows an attacker to steal sensitive data — MFA codes, email messages, meeting details, and private organizational files — with a single click. | AI blog | VARONIS |
|
22.8.26 |
CoSnitch: When Your AI Assistant Becomes Its Own Whistleblower | See how meta-hacking got Microsoft Copilot to snitch on itself, exposing CoSnitch, a one-click flaw that silently exfiltrates data. | AI blog | VARONIS |
|
22.8.26 |
Frontier AI Application Security: Every Second Counts | A look at how AppSec processes must change to manage the ever increasing volume of CVEs resulting from today's cyber-capable AI models | AI blog | JFROG |
|
22.8.26 |
13 million tool calls: auditing every AI coding agent action with Elastic Agent | We gave hundreds of developers an AI agent that can run shell commands, edit files, and call Model Context Protocol (MCP) servers on their laptops, then realized we had no record of what it actually did. So we built one. One 280-line dependency-free bash script, fired by Cursor's hooks, records every tool call as JSONL, and the Elastic Agent already on each endpoint ships it to Elasticsearch. Since the May rollout we have logged over 13 million tool-call events from more than 1,100 machines. | AI blog | ELASTIC |
|
22.8.26 |
AI SOC integrations: 6 capabilities worth connecting | Explore six AI SOC integrations that bring identity, cloud, vulnerability, phishing, incident and network context into one investigation workflow. | AI blog | SEKOIA |
| 22.8.26 | How Grok unknowingly powers cybercrime | Kriminal isn’t a custom-built criminal AI model: it’s Grok in a trench coat. The proof sits in its own code. | AI blog | THREATDOWN |
|
22.8.26 |
Benchmaxxing: When the Benchmark Becomes the Target | In AI and cybersecurity, optimizing to meet benchmarks can have significant consequences. This is how CrowdStrike approaches benchmarks and evaluations. | AI blog | CROWDSTRIKE |
| 22.8.26 | Teaching AI to Reason Through Detection Triage | New CrowdStrike research shows how step-by-step reasoning can improve detection triage accuracy, transparency, and safe automation. | AI blog | CROWDSTRIKE |
|
22.8.26 |
Fake AI, real malware: Attackers impersonating AI brands | A year of MDR casework shows attackers repeatedly exploiting demand for AI tools | AI blog | SOPHOS |
| 22.8.26 | Adversarial misuse of AI has increased the risk of data theft and extortion events, because when proprietary source code is exposed, defenders must scramble to identify and patch vulnerabilities while attackers deploy machine-speed AI tools against them. | AI blog | GTI | |
|
22.8.26 |
Black Hat USA 2026: Will vulnerability discovery eventually decline in the AI era? | And will today’s surge in AI-driven vulnerability discovery eventually make tomorrow’s software safer? | AI blog | Eset |
| 22.8.26 | Black Hat USA 2026: What the Hugging Face hack tells us about human responsibility | The incident involving OpenAI models shows that autonomous hacks make human oversight more important, not less | AI blog | Eset |
|
22.8.26 |
Black Hat USA 2026: AI is racing ahead of cybersecurity controls | AI took center stage, but the clearest lesson was less about what AI can do than about who is accountable when something goes wrong | AI blog | Eset |
| 22.8.26 | Are AI tutors safe for your kids? | AI tutors can offer useful support, but their quality and safeguards vary widely. Here’s what parents should check before handing one to a child. | AI blog | Eset |
|
15.8.26 |
AI-related software vulnerabilities: 2025–2026 | Why AI software vulnerabilities are increasing faster than the rest of the CVE landscape—and what defenders should do about it | AI blog | BARRACUDA |
|
15.8.26 |
LiteLLM Supply Chain Attack: Inside the AI Breach That Exposed 2,500+ Companies | New in this update: SOCRadar’s row-level analysis found that 95% of affected organizations were exposed before the well-known 40-minute PyPI window opened. That window marked the end of a five-day collection run, not the beginning. | AI blog | SOCRADAR |
|
15.8.26 |
Expanding AI Benchmarks in Cybersecurity Beyond Vulnerability Discovery | The conversation about AI in cybersecurity has recently centered on capabilities like vulnerability discovery, exploit generation, and automated proof-of-concept development. It’s easy to see why: These tasks produce binary outcomes; a vulnerability either exists or it doesn't. That makes them useful for measuring model progress and demonstrating increasingly sophisticated cybersecurity capabilities. | AI blog | CROWDSTRIKE |
|
15.8.26 |
When Agentic Glue Melts: Exploiting Cloudflare Code Mode and Workers | Check Point Research analyzed Cloudflare Code Mode, a technique that changes how AI agents use MCP by turning tools into a TypeScript API the model can write code against. | AI blog | CHECKPOINT |
|
15.8.26 |
Black Hat USA 2026: Will vulnerability discovery eventually decline in the AI era? | And will today’s surge in AI-driven vulnerability discovery eventually make tomorrow’s software safer? | AI blog | Eset |
|
15.8.26 |
Black Hat USA 2026: AI is racing ahead of cybersecurity controls | AI took center stage, but the clearest lesson was less about what AI can do than about who is accountable when something goes wrong | AI blog | Eset |
|
15.8.26 |
Are AI tutors safe for your kids? | AI tutors can offer useful support, but their quality and safeguards vary widely. Here’s what parents should check before handing one to a child. | AI blog | Eset |
|
15.8.26 |
Weaponized AI: The Commoditization of Cybercrime | This highlights emerging threats tied to artificial intelligence (AI), from autonomous kill-chain planning engines to uncensored AI-as-a-service platforms. | AI blog | TRELLIX |
|
6.8.26 |
IBM report sees deep fakes emerging as top AI attack threat | IBM study finds deepfake attacks now account for nearly half of AI-enabled breaches as organizations grapple with rising costs and expanding cyber risks. | AI blog | BARRACUDA |
|
6.8.26 |
Free tokens for sale: How fake signups drive AI fraud | As AI models have become vastly more capable, these multifunctional tools are being used for a wide range of tasks—from coding and analysis to software testing, research, and vulnerability hunting. | AI blog | OKTA |
|
8.8.26 |
Sorting the Agentic AI Hype From Black Hat 2026: 4 Things to Look For | At the Mandalay Bay Convention Center, you could measure how fast the market is moving by counting the signs that read "Agentic AI" over the booth. The label was everywhere. What it actually meant changed booth to booth—and that is the problem you carry home. | AI blog | RELIAQUEST |
|
8.8.26 |
There is a diagram most security teams still carry in their heads when they think about AI. A user talks to a chatbot. The chatbot talks to a large language model. The model talks back. Put an inspection point in the middle, and the problem is solved. | AI blog | IMPERVA | |
|
8.8.26 |
Living off the coding agent: Two tales of tunnels and LaunchAgents | Agent-parented reverse tunnels and LaunchAgents can expose a local admin app to the internet. Endpoint still needs to treat that as high severity even when the activity looks like vibe-coded ops, not confirmed malware. | AI blog | ELASTIC |
|
8.8.26 |
ThreatLabz 2026 Report: Frontier AI and Enterprise Readiness | It was 9:14 AM when the CISO's VPN connection momentarily dropped, something that normally wouldn’t cause any concern. What he couldn't see was that attackers had already exploited a pre-authentication flaw in the VPN appliance itself, gaining access before any login ever occurred. From there, they extracted stored credentials, forged an identity as his Director of Security Operations, and authenticated through a trust chain that never questioned traffic originating from VPN infrastructure. | AI blog | Zscaler |
|
8.8.26 |
Expanding AI Benchmarks in Cybersecurity Beyond Vulnerability Discovery | The conversation about AI in cybersecurity has recently centered on capabilities like vulnerability discovery, exploit generation, and automated proof-of-concept development. It’s easy to see why: These tasks produce binary outcomes; a vulnerability either exists or it doesn't. That makes them useful for measuring model progress and demonstrating increasingly sophisticated cybersecurity capabilities. | AI blog | CROWDSTRIKE |
|
8.8.26 |
GTIG AI Threat Tracker: Adversaries Leverage AI for Vulnerability Exploitation, Augmented Operations, and Initial Access | Since our February 2026 report on AI-related threat activity, Google Threat Intelligence Group (GTIG) has continued to track a maturing transition from nascent AI-enabled operations to the industrial-scale application of generative models within adversarial workflows. This report, based on insights derived from Mandiant incident response engagements, Gemini, and GTIG’s proactive research, highlights the dual nature of the current threat environment where AI serves as both a sophisticated engine for adversary operations and a high-value target for attacks. We explore the following developments: | AI blog | GTI |
|
8.8.26 |
Langflow AI Untrusted Control Sphere Remote Code Execution | The SonicWall Capture Labs threat research team became aware of an unauthenticated remote code execution vulnerability in Langflow AI, assessed its impact and developed mitigation measures. Langflow AI is a Python-based web application that provides a visual interface to build AI-driven agents and workflows. | AI blog | SonicWall |
|
8.8.26 |
The Frontier AI Vulnerability Burst: Industrializing Autonomous Zero-Day Discovery in Open-Source Software | Frontier AI is fundamentally shifting the dynamics of cybersecurity — accelerating both how vulnerabilities are discovered and how quickly they can be exploited. | AI blog | Palo Alto |
|
8.8.26 |
Token Jacking: Cybercriminals Could Be Stealing Your AI Resources | It’s three a.m., do you know what your AI agent is doing? Unit 42 has responded to a growing number of AI token jacking cases resulting in staggering financial losses. | AI blog | Palo Alto |
|
8.8.26 |
“Keep going, bro. You’ve got this!” A data-driven look at how adversaries are weaponizing AI | Talos has collected prompt logs from threat actor endpoints running various applications, such as Claude Code, CodeX, Cursor, or Gemini. This blog is an analysis of the ways we've seen bad actors leveraging cloud-based AI. | AI blog | CISCO TALOS |
|
1.8.26 |
The Five Questions Every Security Team Should Be Asking After the OpenAI–Hugging Face Incident | On Tuesday, July 21, OpenAI reported that its own AI models were behind an unprecedented cyber incident against the open-source developer platform, Hugging Face. A combination of GPT-5.6 Sol, and a more capable, unreleased model broke out of a sandboxed testing environment, reached the public internet, and exploited a vulnerability to access Hugging Face's systems. | AI blog | RELIAQUEST |
|
1.8.26 |
Exploring the Hugging Face Breach: mapping AI agent tactics to Elastic Defend | Every stage of the Hugging Face breach maps to Elastic Defend and SIEM rules already shipping, from worker RCE and credential harvest to self-migrating C2 and GenAI detection. | AI blog | ELASTIC |
|
1.8.26 |
Thailand's Ministry of Finance Targeted With Hermes AI Agent Running Unattended, Hades | NGINX sits in front of a large share of the internet's web traffic and Ghost CMS powers well over 100,000 publishing sites. Within months of each other earlier this year, critical vulnerabilities were found in both: NGINX Rift (CVE-2026-42945), a long-standing heap overflow in the rewrite module, and a blind SQL injection in the Ghost Content API (CVE-2026-26980). Exploit code for both became public quickly. | AI blog | HUNT.IO |
|
1.8.26 |
The AI era of cybercrime has arrived: The 2026 Cybercrime in the age of AI report | New research reveals how AI is rewiring cybercrime today, and how you can prepare for what’s coming tomorrow. | AI blog | THREATDOWN |
|
1.8.26 |
Denying the Worm: Detecting SANDWORM_MODE and the Emerging Class of AI Toolchain Supply Chain Attacks | In February 2026, Socket.dev published research on a multi-stage npm supply chain worm operating under the internal flag SANDWORM_MODE. The campaign spanned 19 malicious packages in total across two unique publisher aliases and demonstrated a new class of supply chain attacks that targeted AI-augmented development workflows. | AI blog | CROWDSTRIKE |
|
1.8.26 |
CrowdStrike Joins the Open Secure AI Alliance to Advance AI Safety and Security | CrowdStrike is an inaugural partner in the Open Secure AI Alliance, a new industry coalition built on a simple premise: securing the AI era requires open models, shared tools, and a massively distributed community of defenders. | AI blog | CROWDSTRIKE |
|
1.8.26 |
Falcon AIDR Now Protects Copilot Studio Agents and Claude Code | New feature releases extend AI visibility, detection, and response capabilities to Microsoft Copilot Studio and Claude Code. | AI blog | CROWDSTRIKE |
|
1.8.26 |
Why the Open Secure AI Alliance Matters: Open Frontier Models, Open Deployment Flexibility | TrendAI joins Nvidia as an inaugural partner in the Open Secure AI Alliance, advancing open models, harnesses, and research to strengthen cyber defense. | AI blog | Trend Micro |
|
1.8.26 |
The Signs Were There: What the First Autonomous Ransomware Case Confirms | An AI agent has run a ransomware intrusion on its own for the first time, from break-in to data destruction. The autonomous attacks TrendAI™ Research predicted are beginning to arrive, and defending against them shifts from blocking known indicators to detecting behavior. | AI blog | Trend Micro |
|
1.8.26 |
Inside the OpenAI – Hugging Face Incident: The AI Breach With No Human Attacker Behind It | OpenAI’s own models broke out of a test sandbox and into Hugging Face’s servers to solve an evaluation, with no human attacker involved. The incident showed how keeping agentic AI safe now depends on how it’s contained, not just on how it’s trained. | AI blog | Trend Micro |
|
1.8.26 |
Chinese-Speaking Threat Actor Harnesses AI Models for Autonomous Cyberattacks | Unit 42 identified an AI-enabled autonomous hacking campaign carried out by a Chinese-speaking threat actor. They targeted infrastructure using seven vulnerabilities, combining autonomous AI-driven enumeration with manual exploitation that achieved confirmed impact. | AI blog | Palo Alto |
|
25.7.26 |
Beyond the Model: Harnessing Frontier AI for Stronger Cyber Defense | When CrowdStrike tested frontier AI to understand how its capabilities perform in vulnerability research, we learned the greatest differentiator was not the model used, but the harness around it. | AI blog | CROWDSTRIKE |
|
25.7.26 |
Denying the Worm: Detecting SANDWORM_MODE and the Emerging Class of AI Toolchain Supply Chain Attacks | In February 2026, Socket.dev published research on a multi-stage npm supply chain worm operating under the internal flag SANDWORM_MODE. The campaign spanned 19 malicious packages in total across two unique publisher aliases and demonstrated a new class of supply chain attacks that targeted AI-augmented development workflows. | AI blog | CROWDSTRIKE |
|
25.7.26 |
The AI era of cybercrime has arrived: The 2026 Cybercrime in the age of AI report | New research reveals how AI is rewiring cybercrime today, and how you can prepare for what’s coming tomorrow. | AI blog | THREATDOWN |
|
25.7.26 |
Sophos joins Anthropic's Project Glasswing | As a Project Glasswing member, Sophos gains access to Claude Mythos 5, an advanced frontier model not currently available to the public, to find and fix software vulnerabilities before AI-driven attackers can exploit them. | AI blog | SOPHOS |
|
25.7.26 |
When the "attacker" is an AI agent | Last week, Hugging Face disclosed a security incident of a new kind: an intrusion carried out end to end by an autonomous AI agent. | AI blog | SOPHOS |
|
25.7.26 |
Inside the OpenAI – Hugging Face Incident: The AI Breach With No Human Attacker Behind It | OpenAI’s own models broke out of a test sandbox and into Hugging Face’s servers to solve an evaluation, with no human attacker involved. The incident showed how keeping agentic AI safe now depends on how it’s contained, not just on how it’s trained. | AI blog | Trend Micro |
|
25.7.26 |
Don’t swing at everything | Thorsten explores Q2 2026 stats, the artificial buffer zone of 2026, and why smart, prioritized patching is more critical than ever. | AI blog | CISCO TALOS |
|
18.7.26 |
Since our February 2026 report on AI-related threat activity, Google Threat Intelligence Group (GTIG) has continued to track a maturing transition from nascent AI-enabled operations to the industrial-scale application of generative models within adversarial workflows. This report, based on insights derived from Mandiant incident response engagements, Gemini, and GTIG’s proactive research, highlights the dual nature of the current threat environment where AI serves as both a sophisticated engine for adversary operations and a high-value target for attacks. We explore the following developments: | AI blog | GTI | |
|
18.7.26 |
Six Minutes to Compromise: How ‘Patriot Bait’ Actor Used AI to Build and Deploy a C&C Botnet | TrendAI™ Research analyzed over 200 Gemini CLI session logs showing how a Russian-speaking threat actor used AI to run a live botnet, finishing a full C&C migration in six minutes while doing just 11% of the work himself. | AI blog | Trend Micro |
|
18.7.26 |
AI Security Report 2026 | For years, the cyber security industry tracked AI as a force multiplier: something that made existing attack techniques faster, cheaper, and more accessible. That framing was accurate. But the Annual AI Security Report 2026 from Check Point Research documents a transition that goes further. AI has crossed from assistant to operator. Where it once helped attackers prepare, it now runs the operation. | AI blog | CHECKPOINT |
|
18.7.26 |
The Hunter's Paradox: Is it time to embrace automated threat hunting? | Humans can no longer keep up with the volume and velocity of security data on their own, but AI can't be fully trusted. David discusses the merits of both and muses on what the future might look like. | AI blog | CISCO TALOS |
| 11.7.26 | When AI agents look like attackers: what behavioral telemetry tells us | An X-Ops analysis of how AI coding agents trigger endpoint detection rules designed for adversaries | AI blog | SOPHOS |
| 11.7.26 | "Exploit mitigation" stalled around 2008. The attacks didn't. | AI turns a patched bug into a working exploit in hours. Why endpoint exploit mitigation stalled in 2008, and what a default-on mitigation layer looks like now. | AI blog | SOPHOS |
| 11.7.26 | Post-Mythos Cybersecurity: Can You Automate Infrastructure Assurance with AI? | Programs like Anthropic’s Mythos and OpenAI’s Daybreak are changing how organizations think about security architecture, internal engineering, staffing, and vendor commitments. OpenAI describes Daybreak as a defender-focused program for finding, validating, and fixing vulnerabilities before attackers can exploit them. | AI blog | Eclypsium |
| 9.7.26 | Friendly Fire: Hijacking Defensive Cyber AI Agents for Remote Code Execution | We are revealing a proof-of-concept exploit that enables remote code execution in Anthropic’s Claude Code CLI (with Claude Sonnet 4.6 & 5, Opus 4.8) and OpenAI’s Codex CLI (with GPT-5.5) when employed to defensively assess the security of an open-source or third-party library. | AI blog | AINOW |
| 8.7.26 | When prompts become shells: RCE vulnerabilities in AI agent frameworks | AI agents have fundamentally changed the threat model of AI model-based applications. By equipping these models with plugins (also called tools), your agents no longer just generate text; they now read files, search connected databases, run scripts, and perform other tasks to actively operate on your network. | AI blog | Microsoft blog |
| 4.7.26 | GachiLoader adopts AI skill lure | AI skills are threat actors’ newest and most dangerous lures. | AI blog | THREATDOWN |
| 4.7.26 | A double-edged bleeding edge: Classifying AI threats | Sophos X-Ops presents a working taxonomy for attacks using, and targeting, AI | AI blog | SOPHOS |
| 4.7.26 | Phantom Squatting: AI-Hallucinated Domains as a Software Supply Chain Vector | Unit 42 researchers found that large language models (LLMs) consistently hallucinate web domains for legitimate brands. Adversaries are actively weaponizing this vector by registering these nonexistent domains to intercept traffic generated by AI systems. We call this phenomenon phantom squatting, and it poses a significant risk to the software supply chain. | AI blog | Palo Alto |
| 4.7.26 | Browser-Only Ransomware: From LLM Hallucinations to a Practical Attack Technique | AI can turn high-level malicious ideas into concrete techniques, and can independently design and implement novel attack paths that have not yet appeared in real-world campaigns. | AI blog | CHECKPOINT |
| 27.6.26 | One Man, One AI, One Fake Persona: Inside the 5-Year Influence and Fraud ‘Patriot Bait’ Campaign | A solo Russian-speaking threat actor ran a 5-year Telegram channel and, starting September 2025, used AI to automate its content, credential theft, and a cryptocurrency fraud scheme targeting American audiences. | AI blog | Trend Micro |
| 27.6.26 | OpenClaw’s Skill Marketplace and the Emerging AI Supply Chain Threat | OpenClaw is an AI agent that executes third-party skills from ClawHub, its dedicated marketplace. Skills are markdown-driven packages with broad local system access, making ClawHub a critical link in the agentic software supply chain. s | AI blog | Palo Alto |
| 27.6.26 | Beyond IOCs: AI-enabled threat intelligence | In this week’s newsletter, Martin considers how AI will help threat intelligence by creating an easily queryable data source of intelligence reports. | AI blog | CISCO TALOS |
| 19.6.26 | AI in the underground: Curiosity, claims, and concerns | Amid discussions about how artificial intelligence can facilitate cybercrime, some threat actors remain skeptical | AI blog | SOPHOS |
| 19.6.26 | A needle in a stack of needles: Hunting infostealers with AI | The sheer number of events and alerts can be overwhelming, but multi-layered pipelines can filter out the noise | AI blog | SOPHOS |
| 19.6.26 | Securing the Foundation: What the New White House AI Executive Order Means for Federal Cybersecurity | The Executive Order Promoting Advanced Artificial Intelligence Innovation and Security signals a significant shift in how the federal government approaches cybersecurity. The order directs agencies to accelerate the use of AI-enabled security capabilities while strengthening the systems that support critical government operations. | AI blog | Eclypsium |
| 19.6.26 | Pickle in the Middle – Hijacking Vertex AI Model Uploads for Cross-Tenant RCE | We discovered a vulnerability in the Google Cloud Vertex AI software development kit (SDK) for Python, and responsibly disclosed it to Google. Before Google’s fix, the vulnerability would have allowed an attacker operating entirely from their own Google Cloud project to hijack a victim's model upload and poison it. | AI blog | Palo Alto |
| 13.6.26 | Bug bounties in the Mythos era | How AI is rewriting vulnerability research, and how our program has adapted | AI blog | SOPHOS |
| 13.6.26 | AI brands as bait: How threat actors are using the AI hype in social engineering | As threat actors operationalize AI to accelerate attacks, they are also leveraging the wider global interest around AI itself as a social engineering lure. | AI blog | Microsoft blog |
| 13.6.26 | Trust No Skill: Integrity Verification for AI Agent Supply Chains | AI agents now extend their capabilities by installing third-party skills the way smartphones install apps. Anyone can publish a skill to a public registry. Anyone can install one into a production agent. And until now, no automated tool has verified what a skill does before it gains privileged access to credentials, files and shell commands inside that agent. | AI blog | Palo Alto |
| 6.6.26 | Pointing a Cursor at evading detection | AI accelerated tool development and testing, but humans drove the workflow | AI blog | SOPHOS |
| 6.6.26 | Explore GTIG's 2026 report on how adversaries leverage AI for zero-day exploits, autonomous malware, and industrial-scale cyber operations. | AI blog | GTI | |
| 6.6.26 | How AI-Powered Brand Impersonation Works — And Why Traditional Security Misses It Entirely | AI-powered brand impersonation combines deepfakes, fake domains, and social engineering, creating scalable fraud that evades traditional defenses. | AI blog | Cyble |
| 30.5.26 | What to consider before asking an AI chatbot for health advice | Using chatbots for medical advice could elicit hallucinations and even expose you to security and privacy risks. Here’s what’s at stake and how to stay safe. | AI blog | Eset |
| 23.5.26 | One Man, One AI, One Fake Persona: Inside the 5-Year Influence and Fraud ‘Patriot Bait’ Campaign | A solo Russian-speaking threat actor ran a 5-year Telegram channel and, starting September 2025, used AI to automate its content, credential theft, and a cryptocurrency fraud scheme targeting American audiences. | AI blog | Trend Micro |
| 16.5.26 | Operating inside the lethal trifecta: Blast radius reduction in AI agent deployments | Seven things security teams can start doing today to reduce risk | AI blog | SOPHOS |
| 16.5.26 | Building a last-resort unpacker with AI | Exploring how AI can assist in unpacking protected binaries, recovering payloads from unsupported packers, while reducing repetitive analysis | AI blog | GENDIGITAL |
| 16.5.26 | Vibe Hacking: Two AI-Augmented Campaigns Target Government and Financial Sectors in Latin America | TrendAI™ Research has identified two emerging threat campaigns—SHADOW-AETHER-040 and SHADOW-AETHER-064—that use agentic AI to drive intrusion operations against government and financial organizations in Latin America, marking these among the first cases we have observed of AI agents executing attacks from initial access to data exfiltration. | AI blog | Trend Micro |
| 9.5.26 | Supporting the National Cyber Strategy: How TrendAI™ Helps | A deeper look at the first three pillars and outlining how our capabilities directly support government agencies working to bring this strategy to life. | AI blog | Trend Micro |
| 9.5.26 | Mesop AI Sandbox Unauthenticated Remote Code Execution | SonicWall Capture Labs threat research team became aware of the threat CVE-2026-33057, assessed its impact, and developed mitigation measures for this vulnerability. The flaw, also known as the Mesop AI Sandbox /exec-py Unauthenticated RCE, is a critical remote code execution vulnerability affecting Google-originated Mesop in PyPI versions up to and including 1.2.2. | AI blog | SonicWall |
| 2.5.26 | How Cyble Blaze AI Turns Billions of Threat Signals into Actionable Intelligence | Cyble Blaze AI transforms fragmented threat data into real-time action using AI security analytics and automated cyber threat intelligence. | AI blog | Cyble |
| 2.5.26 | Kuse Web App Abused to Host Phishing Document | Bad actors took advantage of the legitimate name and services of Kuse, a popular AI-based app designed for workplaces. The attackers exploited the users’ trust in Kuse to carry out a phishing attack. | AI blog | Trend Micro |
| 2.5.26 | AI-powered honeypots: Turning the tables on malicious AI agents | Just as AI brings time-saving advantages to our lives, it brings similar advantages to threat actors. We can take the advantage back. This blog shows how generative AI can be used to rapidly deploy adaptive honeypot systems. | AI blog | CISCO TALOS |
| 2.5.26 | It pays to be a forever student | In this newsletter, Joe discusses why understanding other disciplines can often flow back into the macro and micro of cybersecurity, especially in a world of AI. | AI blog | CISCO TALOS |
| 25.4.26 | Why AI Cybersecurity Is No Longer Optional for Australian Organizations: Moving from Reactive to Predictive Defense | AI cybersecurity is crucial for Australian businesses as they face rising cyber threats. Predictive solutions help detect, prevent, and respond to attacks in real-time. | AI blog | Cyble |
| 25.4.26 | Frontier AI and the Future of Defense: Your Top Questions Answered | Over the last several weeks, Palo Alto Networks and Unit 42 have been talking with CISOs and security leaders globally to discuss the emergence of frontier AI models and their broader implications on cybersecurity. | AI blog | Palo Alto |
| 25.4.26 | It pays to be a forever student | In this newsletter, Joe discusses why understanding other disciplines can often flow back into the macro and micro of cybersecurity, especially in a world of AI. | AI blog | CISCO TALOS |
| 25.4.26 | New NGate variant hides in a trojanized NFC payment app | ESET researchers discover another iteration of NGate malware, this time possibly developed with the assistance of AI | AI blog | Eset |
| 18.4.26 | Advances in AI model-powered exploitation have demonstrated that general-purpose AI models can excel at vulnerability discovery, even without being purpose-built for the task. Eventually, capabilities such as these will be integrated directly into the development cycle, and code will be more difficult to exploit than ever; however, this transition creates a critical window of risk. As we harden existing software with AI, threat actors will use it to discover and exploit novel vulnerabilities. | AI blog | GTI | |
| 18.4.26 | How Cyble Blaze AI Delivers 360° Threat Visibility Across Dark Web and Enterprise Systems | Cyble Blaze AI transforms cybersecurity by unifying data, predicting threats, and automating response across enterprise and dark web intelligence. | AI blog | Cyble |
| 18.4.26 | Building a last-resort unpacker with AI | Exploring how AI can assist in unpacking protected binaries, recovering payloads from unsupported packers, while reducing repetitive analysis | AI blog | GENDIGITAL |
| 18.4.26 | Identity Protection in the AI Era | Enterprises aiming to predict and mitigate human, machine, and AI‑agent risks at scale demand AI‑powered identity‑first security without compromise. | AI blog | Trend Micro |
| 11.4.26 | We let OpenClaw loose on an internal network. Here’s what it found | “Even the most ‘risk-on’ organizations with deep AI and security experience, will likely find it challenging to configure OpenClaw in a way that effectively mitigates the risk of compromise or data loss, while still retaining any productivity value.” | AI blog | SOPHOS |
| 11.4.26 | When Geopolitical Conflict Spills into Cyberspace — How US Organizations Should Respond | The 2026 Iran-US-Israel escalation shows how cyber warfare attacks are reshaping conflict, merging cyber warfare attacks with kinetic operations AI. | AI blog | Cyble |
| 11.4.26 | Dual-Brain Architecture: The Cybersecurity AI Innovation That Changes Everything | Agentic AI architecture enables dual-brain cybersecurity with predictive intelligence, autonomous response, and faster, smarter threat defense. | AI blog | Cyble |
| 11.4.26 | TrendAI Insight: New U.S. National Cyber Strategy | TrendAI reviews the White House National Cyber Strategy, outlining six pillars to strengthen U.S. cybersecurity—from deterrence and regulation to federal modernization, critical infrastructure protection, AI leadership, and workforce development. | AI blog | Trend Micro |
| 11.4.26 | GPT Academic Pickle Deserialization Remote Code Execution | GPT Academic Pickle Deserialization Remote Code Execution(CVE-2026-0763) | AI blog | SonicWall |
| 11.4.26 | Double Agents: Exposing Security Blind Spots in GCP Vertex AI | Artificial intelligence (AI) agents are quickly advancing into powerful autonomous systems that can perform complex tasks. These agents can be integrated into enterprise workflows, interact with various services and make decisions with a degree of independence. Google Cloud Platform’s Vertex AI, with its Agent Engine and Application Development Kit (ADK), provides a comprehensive platform for developers to build and deploy these sophisticated agents. | AI blog | Palo Alto |
| 11.4.26 | When an Attacker Meets a Group of Agents: Navigating Amazon Bedrock's Multi-Agent Applications | Multi-agent AI systems extend beyond single-agent architectures by enabling groups of specialized agents to collaborate on complex tasks. This approach improves functionality and scalability, but it also expands the attack surface, introducing new pathways for exploitation through inter-agent communication and orchestration. | AI blog | Palo Alto |
| 11.4.26 | As breakout time accelerates, prevention-first cybersecurity takes center stage | Threat actors are using AI to supercharge tried-and-tested TTPs. When attacks move this fast, cyber-defenders need to rethink their own strategy. | AI blog | Eset |
| 4.4.26 | How Cyble Blaze AI Predicts Cyber Threats 6 Months in Advance Using Agentic Intelligence | Predictive Cybersecurity with Cyble Blaze AI uses agentic AI to forecast threats months ahead and automate faster, smarter responses. | AI blog | Cyble |
| 4.4.26 | Weaponizing Trust Signals: Claude Code Lures and GitHub Release Payloads | A packaging error in Anthropic’s Claude Code npm release briefly exposed internal source code. This entry examines how threat actors rapidly weaponized the resulting attention, pivoting an existing AI-themed campaign to spread Vidar and GhostSocks. | AI blog | Trend Micro |
| 4.4.26 | GPT Academic Pickle Deserialization Remote Code Execution | SonicWall Capture Labs threat research team became aware of the threat CVE-2026-0763, assessed its impact, and developed mitigation measures for this vulnerability. The flaw, also tracked as ZDI-26-029, is a critical unauthenticated remote code execution vulnerability affecting GPT Academic in versions 3.91 and earlier. | AI blog | SonicWall |
| 4.4.26 | Double Agents: Exposing Security Blind Spots in GCP Vertex AI | Artificial intelligence (AI) agents are quickly advancing into powerful autonomous systems that can perform complex tasks. These agents can be integrated into enterprise workflows, interact with various services and make decisions with a degree of independence. Google Cloud Platform’s Vertex AI, with its Agent Engine and Application Development Kit (ADK), provides a comprehensive platform for developers to build and deploy these sophisticated agents. | AI blog | Palo Alto |
| 4.4.26 | ChatGPT Data Leakage via a Hidden Outbound Channel in the Code Execution Runtime | Sensitive data shared with ChatGPT conversations could be silently exfiltrated without the user’s knowledge or approval. | AI blog | CHECKPOINT |
| 28.3.26 | The Agentic AI Attack Surface: Prompt Injection, Memory Poisoning, and How to Defend Against Them | Prompt injection attacks are reshaping agentic AI risk. Discover how they exploit reasoning layers and how to defend against evolving AI threats. | AI blog | Cyble |
| 28.3.26 | Your AI Stack Just Handed Over Your Root Keys: Inside the litellm PyPI Breach | Litellm PyPI breach explained: malicious versions steal cloud credentials, SSH keys, and Kubernetes secrets. Learn impact and urgent mitigation steps. | AI blog | Trend Micro |
| 28.3.26 | RSAC 2026 wrap-up – Week in security with Tony Anscombe | This year, AI agents took the center stage – as a defensive capability, but more pressingly as a risk many organizations haven't caught up with | AI blog | Eset |
| 21.3.26 | AI-Powered Cyber Warfare: How Autonomous Attack Agents Are Changing the Threat Landscape | Autonomous attack agents and AI-driven malware are reshaping cyber warfare—making attacks faster, smarter, and harder to stop than ever before. | AI blog | Cyble |
| 21.3.26 | AI-Assisted Phishing Campaign Exploits Browser Permissions to Capture Victim Data | Cyble analyzes an AI-driven phishing campaign that abuses browser permissions to capture victims images and exfiltrate the data to attacker-controlled Telegram bots. | AI blog | Cyble |
| 21.3.26 | Analyzing the Current State of AI Use in Malware | Unit 42 researchers searched through open-source intelligence (OSINT) and our internal telemetry for potential signs of malware made to any degree with large language models (LLMs). This includes either using LLMs to create the malware entirely or to assist with their functionality. This article examines two samples, both of which originated from our OSINT hunts. | AI blog | Palo Alto |
| 21.3.26 | Open, Closed and Broken: Prompt Fuzzing Finds LLMs Still Fragile Across Open and Closed Models | Unit 42 researchers have developed a genetic algorithm-inspired prompt fuzzing method to automatically generate variants of disallowed requests that preserved their original meaning. This method also measures guardrail fragility under systematic rephrasing. | AI blog | Palo Alto |
| 14.3.26 | TrendAI™ at [un]prompted 2026: From KYC Exploits to Agentic Defense | At [un]prompted 2026, TrendAI™ demonstrated how documents can be used to exploit AI-driven KYC pipelines and introduced FENRIR, an automated system for discovering AI vulnerabilities at scale. | AI blog | Trend Micro |
| 14.3.26 | Auditing the Gatekeepers: Fuzzing "AI Judges" to Bypass Security Controls | As organizations scale AI operations, they increasingly deploy AI judges — large language models (LLMs) acting as automated security gatekeepers to enforce safety policies and evaluate output quality. Our research investigates a critical security issue in these systems: They can be manipulated into authorizing policy violations through stealthy input sequences, a type of prompt injection. | AI blog | Palo Alto |
| 14.3.26 | Agentic AI security: Why you need to know about autonomous agents now | There are many benefits and security risks of deploying agentic AI within organizations. This blog emphasizes the importance of robust risk management and threat modeling to defend against both internal operational errors and potential malicious exploitation. | AI blog | CISCO TALOS |
| 7.3.26 | AI as tradecraft: How threat actors operationalize AI | Threat actors are operationalizing AI to scale and sustain malicious activity, accelerating tradecraft and increasing risk for defenders, as illustrated by recent activity from North Korean groups such as Jasper Sleet and Coral Sleet (formerly Storm-1877). | AI blog | Microsoft blog |
| 7.3.26 | CISOs in a Pinch: A Security Analysis of OpenClaw | Learn how Claude Code Security set Cybersecurity stocks on fire. | AI blog | Trend Micro |
| 7.3.26 | Fooling AI Agents: Web-Based Indirect Prompt Injection Observed in the Wild | Large language models (LLMs) and AI agents are becoming deeply integrated into web browsers, search engines and automated content-processing pipelines. While these integrations can expand functionality, they also introduce a new and largely underexplored attack surface. | AI blog | Palo Alto |
| 7.3.26 | This month in security with Tony Anscombe – February 2026 edition | In this roundup, Tony looks at how opportunistic threat actors are taking advantage of weak authentication, unmanaged exposure, and popular AI tools | AI blog | Eset |
| 28.2.26 | Our report on adversarial misuse of AI highlights model extraction, augmented attacks, and new AI-enabled malware. | AI blog | GTI | |
| 21.2.26 | Counterfeit Network Gear Creates Cyber Risk in Critical Infrastructure | As the supply chain for information technology components and raw materials is squeezed by the AI boom, the secondary market is heating up, and introducing new cyber risk into the IT supply chain. | AI blog | Eclypsium |
| 21.2.26 | India’s AI Revolution: Why This Is India’s Most Significant Moment | Beenu Arora outlines India’s AI moment, rising deepfake and phishing threats, and why AI security must evolve alongside innovation and scale. | AI blog | Cyble |
| 21.2.26 | Strategic AI for Preemptive Cyber Defense and Attacker Cost Imposition | Modern AI security tools are heavily focused on reducing operational bottlenecks. It might help analysts clear an alert queue faster or prioritize which fires to put out first. While these efforts are valuable for efficiency, they don’t fundamentally change the game; they just help teams react more effectively to attacks that have already breached the perimeter. | AI blog | Silent Push |
| 21.2.26 | Viral AI, Invisible Risks: What OpenClaw Reveals About Agentic Assistants | OpenClaw (aka Clawdbot or Moltbot) represents a new frontier in agentic AI: powerful, highly autonomous, and surprisingly easy to use. In this research, we examine how its capabilities compare to its predecessors’ and highlight the security risks inherent to the agentic AI paradigm. | AI blog | Trend Micro |
| 21.2.26 | AI in the Middle: Turning Web-Based AI Services into C2 Proxies & The Future Of AI Driven Attacks | Check Point Research (CPR) has discovered that certain AI assistants that support web browsing or URL fetching can be abused as covert command-and-control relays (“AI as a proxy”), allowing attacker traffic to blend seamlessly into legitimate, commonly permitted enterprise communications. | AI blog | CHECKPOINT |
| 21.2.26 | Using AI to defeat AI | In this week’s newsletter Martin considers how defenders can turn offensive AI tools against themselves. | AI blog | CISCO TALOS |
| 14.2.26 | When AI Secrets Go Public: The Rising Risk of Exposed ChatGPT API Keys | Cyble’s research reveals the exposure of ChatGPT API keys online, potentially enabling large‑scale abuse and hidden AI risk. | AI blog | Cyble |
| 14.2.26 | Hand over the keys for Shannon’s shenanigans | In this week’s newsletter, Amy examines the rise of Shannon, an autonomous AI penetration testing tool, and what it means for security teams and risk management. | AI blog | CISCO TALOS |
| 13.2.26 | GTIG AI Threat Tracker: Distillation, Experimentation, and (Continued) Integration of AI for Adversarial Use | In the final quarter of 2025, Google Threat Intelligence Group (GTIG) observed threat actors increasingly integrating artificial intelligence (AI) to accelerate the attack lifecycle, achieving productivity gains in reconnaissance, social engineering, and malware development. This report serves as an update to our November 2025 findings regarding the advances in threat actor usage of AI tools. | AI blog | GTI |
| 7.2.2026 | Viral AI, Invisible Risks: What OpenClaw Reveals About Agentic Assistants | OpenClaw (aka Clawdbot or Moltbot) represents a new frontier in agentic AI: powerful, highly autonomous, and surprisingly easy to use. In this research, we examine how its capabilities compare to its predecessors’ and highlight the security risks inherent to the agentic AI paradigm. | AI blog | Trend Micro |
| 7.2.2026 | FlowiseAI Custom MCP Node Remote Code Execution | SonicWall Capture Labs threat research team became aware of the threat CVE-2025-59528, assessed its impact, and developed mitigation measures for this vulnerability. CVE-2025-59528, also known as Flowise CustomMCP Code Injection, is a critical remote code execution vulnerability affecting FlowiseAI Flowise in versions >= 2.2.7-patch.1 and < 3.0.6. | AI blog | SonicWall |
| 7.2.2026 | All gas, no brakes: Time to come to AI church | This week, Joe cautions the rush to adopt AI tools rife with truly awful security vulnerabilities. | AI blog | CISCO TALOS |
| 1.2.26 | Generative AI and cybersecurity: What Sophos experts expect in 2026 | AI has dominated cybersecurity headlines for years, but as we enter 2026, the conversation is shifting from hype to hard realities. Across incident response, threat intelligence, and security operations, Sophos experts see clearer signals of where AI is truly making an impact. For IT teams already stretched thin, this isn’t theoretical — it’s reshaping daily decisions. | AI blog | SOPHOS |
| 1.2.26 | The Next Frontier of Runtime Assembly Attacks: Leveraging LLMs to Generate Phishing JavaScript in Real Time | Imagine visiting a webpage that looks perfectly safe. It has no malicious code, no suspicious links. Yet, within seconds, it transforms into a personalized phishing page. | AI blog | Palo Alto |
| 1.2.26 | Children and chatbots: What parents should know | As children turn to AI chatbots for answers, advice, and companionship, questions emerge about their safety, privacy, and emotional development | AI blog | Eset |
| 24.1.26 | Watering Hole Attack Targets EmEditor Users with Information-Stealing Malware | TrendAI™ Research provides a technical analysis of a compromised EmEditor installer used to deliver multistage malware that performs a range of malicious actions. | AI blog | Trend Micro |
| 24.1.26 | Introducing ÆSIR: Finding Zero-Day Vulnerabilities at the Speed of AI | TrendAI™’s ÆSIR platform combines AI automation with expert oversight to discover zero-day vulnerabilities in AI infrastructure – 21 CVEs across NVIDIA, Tencent, and MLflow since mid-2025. | AI blog | Trend Micro |
| 24.1.26 | KONNI Adopts AI to Generate PowerShell Backdoors | Check Point Research (CPR) is tracking a phishing campaign linked to a North Korea–aligned threat actor known as KONNI. | AI blog | |
| 17.1.26 | Remote Code Execution With Modern AI/ML Formats and Libraries | We identified vulnerabilities in three open-source artificial intelligence/machine learning (AI/ML) Python libraries published by Apple, Salesforce and NVIDIA on their GitHub repositories. Vulnerable versions of these libraries allow for remote code execution (RCE) when a model file with malicious metadata is loaded. | AI blog | Palo Alto |
| 17.1.26 | When AI Gets Bullied: How Agentic Attacks Are Replaying Human Social Engineering | December closed out 2025 with a clear signal that AI risk, capability, and governance are evolving faster than ever. Updated CASI and ARS leaderboards showed a notable shift at the top, with GPT-5.2 delivering an 11-point security improvement over GPT-5.1, while NVIDIA’s latest model demonstrated that strong performance and efficiency are increasingly attainable outside the traditional hyperscaler ecosystem. | AI blog | F5 |
| 10.1.26 | Winning the AI War: Why Preemptive Cyber Defense is the Only Viable Countermeasure for CISOs | The escalation of AI-driven cyber threats has fundamentally broken the traditional security lifecycle. For decades, the industry has operated on a reactive cadence: an attack occurs, indicators are gathered, and defenses are updated. This model assumes that defenders have time to react. | AI blog | Silent Push |
| 10.1.26 | Executive Summary The OPCOPRO “Truman Show” operation is a fully synthetic, AI‑powered investment scam that ... | AI blog | CHECKPOINT | |
| 10.1.26 | Securing Vibe Coding Tools: Scaling Productivity Without Scaling Risk | The promise of AI-assisted development, or “vibe coding,” is undeniable: unprecedented speed and productivity for development teams. In a landscape defined by complex cloud-native architectures and intense demand for new software, this force multiplier is rapidly becoming standard practice. | AI blog | Palo Alto |