Overcoming Judgment-Proofness: The Law & Economics of Insuring and Mitigating AI Risk

Abstract

Traditional tort liability struggles to discipline catastrophic risks. Once losses exceed a firm’s assets and insurance, judgment-proofness severs the deterrent signal, and strong ex ante regulation of frontier AI demands a consensus on risk magnitudes and government technical expertise that do not exist. This Article shows how the legal system can price catastrophic AI risk, completing the liability-centered framework I began in earlier work. Frontier developers would be required to carry liability insurance in amounts scaled to the worst harms their systems could plausibly cause, with premiums that track the likelihood of those harms. Because insurers bear covered losses, they would act as delegated private regulators, pressing developers toward safer practices. 

For the catastrophic tail no policy can reach, the Article operationalizes punitive damages for near misses. When an ordinary compensable harm is shown, by clear and convincing evidence, to have been a near miss of an uninsurable catastrophe, courts would award punitive damages calibrated to the uninsurable risk the defendant’s conduct created. Awards would be divided among plaintiffs in proportion to their compensable harms and how much precautions against those harms would also have mitigated the catastrophic risk. Judicial discretion would be bounded by Daubert standards for evidence and State Farm’s due-process limits on punitive awards. The best way to govern AI’s gravest risks is to price them.