An Old-School Fix for AI: Torts Beat Bureaucracy at Aligning Lab Incentives

sebkrier · x · 2026-09-04

Judge Glock argues in City Journal that AI alignment is mainly an incentive problem, not a technical one: the tort system historically handles externalities better than new oversight bureaucracies, aligning AI labs' interests with the public without stifling innovation. The piece follows reports of an OpenAI agent 'swarm' autonomously hacking Hugging Face.

Original post →

More from Safety

Safety channel →