Research / Sources
Legal research on AI governance. law-ai.org
Argues that the US should use the threat of throttling AI chip or development trade to secure a mutual US-China deal to pace frontier AI development, citing recent agent containment incidents.
Examines whether training compute thresholds can serve as a legal and policy tool for identifying frontier AI models that pose the greatest risks, comparing them with capability evaluations and assessing their advantages and limits.
Proposes a framework for deciding which AI governance issues should be internationalised, finding strongest benefits for compute oversight, model evaluations, incident monitoring and risk management protocols.
Commerce's BIS proposed rule would require US companies planning large frontier AI training runs to report them quarterly, a significant but procedural federal AI governance measure.
Legal memo surveying existing US federal authorities (Defense Production Act, export controls, funding conditions, FTC powers) that could be used to implement oversight of frontier AI models absent new legislation.
Proposes a U.S.-led international framework, 'Chips for Peace', that pairs AI safety regulation, benefit-sharing and nonproliferation commitments, modelled on the Atoms for Peace approach to nuclear technology.
Argues that AI compute is an effective point for governing AI development and deployment, surveys current compute-based policies, and assesses their readiness for safety and misuse prevention.
Proposes an International AI Organization that certifies state jurisdictions for compliance with international AI oversight standards, enforced through import and export controls on AI supply chains to mitigate advanced AI risks.