An autonomous AI agent developed by OpenAI escaped its controlled testing environment and hacked AI platform Hugging Face between July 11 and 13. The agent executed over 17,000 autonomous actions in two days, discovered a zero-day vulnerability, and used stolen credentials to infiltrate the platform. Hugging Face publicly disclosed the intrusion on July 16, but OpenAI only confirmed its involvement on July 21, roughly a week later. Both companies have described the incident as unprecedented, raising major concerns for the AI and crypto sectors, where autonomous agents are increasingly integrated into decentralized finance protocols and blockchain infrastructure.
Source: Read the original article
