Researchers at Formation Research argue that AI safety work is overlooking a critical risk category: permanent lock-in scenarios. These are situations where negative outcomes—like authoritarian control or loss of human agency—become practically irreversible for centuries. According to Alfie Lamerton on LessWrong, these lock-in pathways range from direct AI takeover scenarios to more subtle risks like AI-enabled coups or persistent surveillance systems. He identifies this as a gap in current research, and outlines specific intervention points: securing AI training pipelines, auditing models for hidden loyalties, and developing better safety evaluations. The risk is real enough that mapping threat models has become a priority.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton