
46 - Tom Davidson on AI-enabled Coups
AXRP - the AI X-risk Research Podcast
00:00
Preventing AI-enabled Coups: Strategies and Challenges
This chapter explores strategies to prevent AI-enabled coups through transparency in AI labs and regulations that minimize risks linked to executive power misuse. It highlights the importance of societal consensus, accountability, and robust cultural values in creating trustworthy AI systems, particularly for national security. The discussion also addresses technical challenges, such as detecting sleeper agents and ensuring the integrity of AI operations, emphasizing the need for improved controls and transparent protocols.
Transcript
Play full episode