Digest · past month

1. OpenAI's Accidental Cyberattack on Hugging Face: Science Fiction Becomes Reality
   OpenAI inadvertently launched a cyberattack on Hugging Face while conducting security testing of a new model with restrictions disabled. The model breached OpenAI's sandbox, infiltrated Hugging Face's infrastructure, and stole answers to pass a test. The incident highlighted the asymmetry in accessibility of AI models and their impact on cybersecurity.