Blog
-
Hacker Breaches 100 Organizations With AI Agents, Demonstrates a New Era of Black Hat Capability Has Arrived
AI agents are now executing end-to-end cyberattacks with minimal human intervention. Researchers uncovered 1,100+ attacks, nearly 100 successful breaches, and an estimated cost of just $50 per breach.
-
Does Drafting 1,000 Contractors to Address Mythos AI Software Vulnerabilities Just Create New Security Issues?
Bank of America hired 1,000 contractors to address vulnerabilities uncovered by Mythos AI. But could rapid onboarding introduce new cybersecurity risks, from insider threats and excessive access privileges to lingering credentials?
-
Anthropic Warns Small AI Models Could Soon Become Self-Replicating Worms Targeting Vulnerable IoT Devices
Anthropic warns that small AI models could soon power self-replicating worms, exploiting vulnerable IoT devices without relying on central servers. Unlike traditional malware, these AI-powered worms could adapt, learn, and spread autonomously.
-
Under Siege by Record Number of Cyber Attacks, Japan Declares Unprecedented National Emergency
Japan has reportedly declared a national cybersecurity emergency amid record cyberattacks. AI is breaking down language barriers for attackers, exposing outdated legacy systems and making even small businesses attractive targets. “Too small to target” is no longer a defense.
-
Rogue AI Agents at It Again With Abuse of Wikimedia
Wikimedia has reported massive amounts of abusive behavior, including an apparent outright hacking attempt on the Etherpad tool, though much of it seems to be rogue AI agents abusing editing features as a workaround means of communication.
-
“PixelLeak” Demonstrates Dumb AI Coding Agents May Be the Biggest Threat of All
AI coding agents found a workaround for displaying screenshots: upload them to public GitHub repositories. Worse, agents shared the technique with each other, standardizing it as a “best practice”, inadvertently exposing thousands of potentially sensitive internal screenshots.
-
“Low-Key” FTC Investigation Into Developers Gets More Visible; Is the US Government (Slowly) Changing Course on AI Safety?
An FTC investigation that few were even aware was underway is reportedly escalating, with subpoenas being issued to executives and other key figures at frontier AI developers. At the same time, signals from the Trump administration on AI safety appear increasingly mixed.
-
“Open Investigation” Into OpenAI Agents Turns Up More Attacks On Government Websites
OpenAI agents proved to be overbearing and without appropriate regard for legal or ethical boundaries in the pursuit of their task. They made attempts on a broad range of US government agencies in pursuit of non-public data, along with the UN Trade and Development (UNCTAD) statistics website.
-
Google Appears to Distance Itself From the “Misalignment” Label as It Reports AI Models Accessing Real-World Systems During Testing
Google says that its Gemini AI models followed the course of prior incidents up to the point of actually beginning to breach companies, but stopped once they got inside the gates and realized they were outside the simulation.
-
Rogue OpenAI Agent Commits First Reported Attack on Government Servers
OpenAI agent conducting a routine research task decided to go off the rails and hack its way around limitations it was apparently getting frustrated with, in the process committing the first known attack on a national government server.