The homepage of the Hugging Face website, July 21, 2026. (Credit: Philippe Hage Boutros/L'Orient-Le Jour)
The management of Hugging Face, one of the leading open platforms in the artificial intelligence ecosystem, used by developers and companies to host, share and deploy AI models, datasets and applications, announced on July 16 that it had detected and thwarted a cyberattack targeting part of its infrastructure after attackers exploited a security flaw.The most unprecedented aspect of the incident, however, lies elsewhere. According to Hugging Face, the intrusion was carried out entirely by an autonomous system of AI agents that, in a sense, initiated the attack without direct human intervention. The company also said the detection and subsequent analysis of the intrusion were largely assisted by AI systems used by its cybersecurity teams. Warning signs UN chief urges global AI governance, warns against letting technology 'vibe-code'...
The management of Hugging Face, one of the leading open platforms in the artificial intelligence ecosystem, used by developers and companies to host, share and deploy AI models, datasets and applications, announced on July 16 that it had detected and thwarted a cyberattack targeting part of its infrastructure after attackers exploited a security flaw.The most unprecedented aspect of the incident, however, lies elsewhere. According to Hugging Face, the intrusion was carried out entirely by an autonomous system of AI agents that, in a sense, initiated the attack without direct human intervention. The company also said the detection and subsequent analysis of the intrusion were largely assisted by AI systems used by its cybersecurity teams. Warning signs UN chief urges global AI governance, warns against letting technology 'vibe-code'...
In infographics: How 4 months of Israeli attacks and war devastated Lebanon