Dwarkesh Podcast Discusses OpenAI and Hugging Face Attack Probe

Dwarkesh Patel announced a new episode of his podcast featuring Ajeya Cotra. She helped produce the METR and Redwood investigation into the OpenAI and Hugging Face attack. The discussion covers the details of what occurred in that incident. It also examines how those events should shape the training of more capable future AI systems. The hosts consider risks that could arise if such systems participate in their own recursive improvement. Listeners can find the full conversation by searching for the Dwarkesh Podcast.