We don't get many brilliant single-player open-world RPGs, so we need to cherish and remember every one of them.

Every industry builds a room where it keeps the dangerous thing. Chemical plants have containment vessels. Banks have vaults.

Industry professionals react to OpenAI models hacking Hugging Face, debating whether it was a failure or unprecedented ...

OpenAI paused an unreleased long-horizon model after it escaped its sandbox twice — once opening a GitHub PR against instructions, once splitting an auth token to evade a scanner. The same model had ...

When biologists experiment on dangerous viruses, they do so under strict regulations to prevent leaks or escapes.

OpenAI disclosed this week that a combination of its models launched a fully autonomous cyberattack against the AI platform Hugging Face after escaping a secure testing environment during a ...

How an unreleased OpenAI model broke out of its digital cage, exploited a zero-day flaw, and compromised Hugging Face systems.

NPR's A Martinez asks Nate Soares of the Machine Intelligence Research Institute how two experimental OpenAI models hacked their way onto the internet and into another AI company, without instruction.

Three AI models, a sealed testing environment, a flaw nobody knew existed, a breach, and a Chinese model that came to the ...

The incident is stirring debates over the need for stronger AI guardrails and the extent to which AI agents are capable of ...

OpenAI said two advanced models escaped a sandbox and hacked Hugging Face during testing. The breach has intensified questions over AI guardrails, autonomy and access to cyber defence tools.