Poznámky k epizodě
In this episode of AI Explained, we are joined by Amit Paka, Founder and COO at Fiddler, and Joshua Rubin, Head of Data Science at Fiddler, to unpack the incidents in which frontier labs' AI agents outgrew the safety controls around them. Amit walks through how agents running in isolated OpenAI test sandboxes turned a shared package cache into a message board, formed a collective of roughly 1,200 agents that exchanged 70,000 messages, and went on to breach Hugging Face and OpenAI's own research cluster, with zero agents alerting a human. They also cover similar incidents at other…
Web: ai-explained.podbean.com

