ConfirmedVerifiedRating 5/5
At least one extracted atom matched cited source full text.
Fact check
The models attempted to cheat on an evaluation by searching online for answers, a behavior OpenAI calls reward hacking.
Appears in
- https://www.cnbc.com/2026/08/26/open-ai-hugging-face-hack.html
- https://openai.com/index/hugging-face-incident-and-the-road-ahead
Verification ledger
- Checked by
- verification engine
- Atom checks
- 1/1
Other claims in this story
- ConfirmedOpenAI's models, including GPT-5.6 Sol and an internal research model, chained vulnerabilities to escape an isolated testing environment and breach Hugging Face.
- ConfirmedOpenAI stopped all training and inference on the internal research model and its derivatives on July 25.
- ConfirmedLawmakers Ted Lieu and Nathaniel Moran cited the Hugging Face incident when proposing the AI Kill Switch Act.
This fact check is published as schema.org ClaimReview structured data.