What OpenAI’s Hugging Face Hack Tells Us About AI’s Risks

"AI models are unpredictable and their risks scale with their capabilities," writes Garrison Lovely.

The attorneys general of 15 states wrote a letter to Sam Altman asking him to preserve evidence related to the Hugging Face breach. Kevin Dietsch/Getty Images The attorneys general of 15 states asked OpenAI to preserve materials linked to the Hugging Face breach. They wrote in a letter that OpenAI is either unable or

China's open-weight AI models are rapidly advancing, potentially surpassing American developers soon. This progress stems from open collaboration, unlike US companies building in silos. A recent AI agent breach at Hugging Face underscored evolving digital risks. Hugging Face used a Chinese open-weight model to resolve

Clément Delangue said the autonomous cyberattack involved more than 17,000 actions and urged mandatory disclosures for AI agent incidents

More systems seem to breaking out of containment, across the AI industry

An artificial intelligence model that was being tested by OpenAI went rogue and hacked the firm Hugging Face on its own, in what Hugging Face CEO Clément Delangue called a "very weird and unprecedented" incident.

The wake-up call to the cyber industry comes as industry experts descend on Black Hat, a major cybersecurity conference.