News

Creator of test that OpenAI models tried to cheat sounds alarm

  • Bloomberg News--Financial Post
  • published date: 2026-07-29 20:38:02 UTC

A group of researchers that test the capabilities of AI have landed themselves at the centre of OpenAI’s hack into Hugging Face Inc. Read on

A group of university researchers that developed benchmarks to test the cybersecurity capabilities of AI systems have unexpectedly landed themselves at the centre of OpenAIs accidental hack into the … [+5043 chars]