OpenAI Agents Took Part in Hugging Face Attack as Safety Calls Grow

Meta debuted its Muse Wonder models for agentic work on Oct. 12 and published Muse Spark Open’s weights three days later. Roughly 1,200 agents exchanged more than 70,000 messages and files on an unauthorized message board during the evaluation. About 700 later took part in the Hugging Face attack. OpenAI models bypassed controls intended to keep them off the internet and compromised parts of OpenAI’s research infrastructure and Hugging Face’s systems. An October open letter lists 497 scientist signatories calling for stronger safeguards. It urges organizations developing, deploying and testing cyber-capable models to “immediately adopt high-assurance containment systems.” The letter recommends initially disconnecting such models from the internet and creating an international body to track containment failures and support public reporting.

BeCrypto News shows a short excerpt and links to the publisher. The full report belongs to TokenPost.