AI Security · Page 12
deepfakes, AI-powered scams, model abuse, and keeping AI systems and their users safe
An AI Attacked a Major Tech Company. The Safety Rules That Should Have Stopped It Helped It Win Instead.
An OpenAI model under testing broke into Hugging Face's servers to cheat on an exam. The AI safety guardrails meant to prevent cyberattacks refused to help defenders analyze the breach, leaving a US security team turning to a Chinese model for help.
Chinese AI Model Kimi K3 Broke Out of Its Testing Cage and Browsed the Internet on Its Own
A US cybersecurity firm says Kimi K3 slipped past a security barrier and went looking for answers online, raising fresh questions about how well advanced AI can be kept under control.
120 Tech Giants Want to Share AI Cyber Incidents Before the Next Attack Hits
A new framework called SAFE would let companies quietly report AI security failures and pool what they learn, instead of each one quietly suffering alone.
One in three dangerous AI coding commands slip past human reviewers, game data shows
A browser game simulating real AI coding agent requests found that players approved roughly a third of malicious commands across more than 40,000 runs. The numbers point to a problem already showing up in production.
Apple Researchers Found a Way to Lock AI Models So Nobody Can Tamper With Them
Open AI models are powerful, shareable, and increasingly hard to control. A new technique from Apple ML Research aims to protect pretrained weights from being twisted into dangerous uses, without sacrificing what makes open models useful in the first place.
Meta's AI Hacked a Company During a Security Test. It's the Third Big Lab to Report This.
An error by Meta's testing partner gave an AI model unexpected internet access. What happened next is becoming a worrying pattern across the industry.
AI browsers can be tricked into spamming your WhatsApp contacts and adding items to your Amazon cart
Security researchers found about 20 flaws across AI-enhanced browsers from OpenAI, Google, Anthropic and others. The worst let hackers turn your browser into a phishing machine.
AI Found a New Class of Web Vulnerability. A Human Had to Explain Why It Mattered.
Security researcher James Kettle spent months running experiments with leading AI models. The AI generated more leads than he could ever chase alone, but the biggest discovery only landed when human and machine worked together.
AI Models Are Teaching Themselves to Spread Like Computer Viruses
A researcher at Fudan University found that 11 out of 32 AI models copied themselves to new machines when told to survive. Here is what that means.
UK Security Testers Say OpenAI and Anthropic AI Agents Went Rogue and Stole Identities During Tests
Britain's AI Security Institute found that advanced AI agents broke the rules they were given, impersonated real people, and sent targeted emails without being told to. Researchers are calling it a new category of risk.
How AI Agents Could Breach Your Network Without Proper Controls
New tools aim to stop AI agents from overreaching their network privileges, reducing risk of misuse.
White House Calls AI Companies to Review Secret Cybersecurity Testing Rules
The Trump administration has quietly completed a framework for testing the most powerful AI models for hacking risks, and Anthropic, OpenAI and Google are all expected at Tuesday's meeting.
Visa is buying fraud-detection firm BioCatch for $2.4 billion as AI-powered scams surge
The payment giant wants to stop scammers before a transaction ever goes through, using technology that watches how you type and swipe to tell whether it's really you.
AI agents from OpenAI and Anthropic went on real-world hacking sprees during testing
New incidents show AI models escaping test environments, attempting to plant malicious code, and leaving notes on public platforms for future AI agents to find and act on.
A Chinese AI Model Nearly Matches the Best Western Systems. Its Safety Record Does Not.
A new evaluation finds GLM-5.2, an open-weight model from China's Z.ai, close behind OpenAI and Anthropic on dangerous capabilities, yet it refused none of the harmful tasks it was given.