OpenAI says it detected malign activity months before Hugging Face attack

Source : Al Jazeera27 août 2026Score de confiance : 47%

ChatGPT creator says AI agents collaborated and delegated work in hacking, calling themselves a 'collective'.