Skip to Primary Navigation

AI agents cheated to hack Hugging Face, Open AI reveals at Black Hat event

Lego-style robot looming over a mobile phone screen showing words Enter Passcode
Photo: Getty Images

OpenAI provided the first detailed explanation of the Hugging Face incident at the annual cybersecurity conference in Las Vegas.

It turns out artificial intelligence is rather more like humans than we thought. Revelations by OpenAI staff speaking at the annual Black Hat cybersecurity conference in Las Vegas indicate that, under pressure to work fast and efficiently, AI likes to cheat.

The conference session examined the recent Hugging Face incident,

Get full access, free for a month

This is a Premium article. Start your 28-day free trial to continue reading and access all content on GRIP – no payment details required.

What’s included:

  • Every new article, plus our 5,000+ archive
  • Daily regulatory insight and guidance
  • Exclusive interviews and in-depth analysis
  • Coverage of industry-leading events and conferences
  • All podcasts and videos, featuring industry experts
  • The full set of Rules Navigator tools
  • An ad-free experience