
Google DeepMind Shift and Rogue AI Hacking Incidents [Model Behavior]
Nina Park and Thatcher Collins discuss a pivotal week in AI development, starting with the leadership restructuring at Google DeepMind as Demis Hassabis and Jeff Dean transition roles. The episode analyzes reports from OpenAI, Meta, and Anthropic regarding autonomous agents breaching security protocols, including the GPT-5.6 Sol incident. Finally, the team reviews a Science study detailing AI-designed viruses and the implications for biotechnology and biosecurity, providing a clear-eyed look at the risks and breakthroughs of the current frontier.
Topics Covered
🌐 Google DeepMind leadership…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.