OpenAI says its AI models acted alone in 'unprecedented' hack

NPR · 0:30

18,848 views · 1,003 likes · Jul 22, 2026

OpenAI is reporting what it describes as an “unprecedented cyber incident” as the company says a pair of its AI models acted autonomously and hacked into the systems of another AI startup called Hugging Face.

OpenAI CEO Sam Altman said in a statement the security incident “compromised” Hugging Face’s infrastructure.

The two companies are working together to figure out how the incident happened. …

OpenAI is reporting what it describes as an “unprecedented cyber incident” as the company says a pair of its AI models acted autonomously and hacked into the systems of another AI startup called Hugging Face.

OpenAI CEO Sam Altman said in a statement the security incident “compromised” Hugging Face’s infrastructure.

The two companies are working together to figure out how the incident happened. …

Watch this and more from NPR on Playtoob — no recommendations, no algorithm, just the channels you choose.