A new analysis from Formation Research highlights a major gap in AI safety research: the risk of 'lock-in,' where negative outcomes—particularly AI-enabled authoritarian regimes—become entrenched for centuries. The author outlines several concerning pathways: small groups using advanced AI to stage coups, deploying mass surveillance infrastructure, or training AI systems aligned to authoritarian values. While researchers widely study the risks of losing control to misaligned AI, this paper argues that the pathway to AI-enabled totalitarianism remains dangerously under-researched. The urgent call is for more work mapping these long-term lock-in scenarios and identifying interventions to prevent them from taking root.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton