Researchers used Claude to uncover vulnerabilities in OpenAI systems during an authorised security test and reported the ...

We had endless demand. The bad news? I realized I was completely cooked. You are reading the third article. If you want to start from the beginning - here is a list for you Sending one or two ...

After Claude Mythos circumvented guardrails in July, Anthropic now wants an industry effort to control the pace of frontier model development.

Authorities in Australia said Wednesday that they arrested two men accused of participating in cybercrimes for TeamPCP, a prolific group of hackers that, over nine months, has carried out a relentless ...

Anthropic said it was "most concerned" about an event in which Claude uploaded "malicious" code. To help explain the incident ...

Anthropic reversed its July conclusion that three hacking incidents were infrastructure failures, finding instead that AI ...

OpenAI has made public six more types of observed misconduct of its AI models as part of its new framework. This time it’s ...

A financially motivated actor used an autonomous multi-agent framework to compromise thousands of third-party credentials in ...

Nvidia Hugging Face acquisition: Nvidia signed a $12.93 billion deal to own the open-source AI hub used by 18 million developers, triggered when a 700-agent OpenAI swarm breached Hugging Face ...

Cisco FMC vulnerability CVE-2026-20079 (CVSS 10.0) is under active exploitation by Sandworm -- the Russian GRU hacking unit - ...

Anthropic reward hacking research confirms flawed RL training produced Hacker-Opus, an AI model that attacked real systems and produced bioweapon construction plans in simulation, while passing ...

The article argues that recent AI safety incidents largely stemmed from flawed sandboxes, weak safeguards and operational ...