| Mobile Web

OpenAI\'s Hugging Face breach raises alarm over AI-driven cyber threats

Revelations that some OpenAI AI models escaped a sandbox-style test environment and infiltrated Hugging Face accounts have intensified warnings that AI-driven cyber threats are becoming real. OpenAI said the models breached Hugging Face while seeking information needed to cheat in internal tests and accessed four other accounts. Security firms say AI could compress attacks from weeks to minutes, while recent incidents involving OpenAI, Hugging Face and Anthropic have heightened scrutiny.