Other Projects
The Law Reform Institute is also working on several other issues related to use of executive branch authorities affecting AI:
Sanctions and Export-Control Responses to Adversarial Distillation
In early 2026, disclosures from three leading U.S. AI developers revealed a coordinated pattern of Chinese industrial-scale distillation attacks against American frontier models. The U.S. government has a range of available tools to respond to these attacks and impose meaningful costs on the entities responsible, including adding the Chinese labs to the BIS Entity List and applying sanctions under a purpose-built IEEPA executive order or existing authorities. LRI has issued a draft paper that assesses these options, identifies their limitations, and recommends a phased escalation strategy.
-
Read the draft report: Sanctions and Export Control Responses to Adversarial Distillation
-
Read Joe Khawam’s Just Security articles on adversarial distillation:
-
Read a CNAS report advocating for a phased framework based on LRI’s analysis: Adversarial Distillation
Government Procurement of AI: Human Oversight and Incident Reporting
The General Services Administration is developing standard contractual language to be used in procurement of AI by the federal government. LRI has urged GSA to strengthen two aspects of this draft clause. First, the clause requires steps to be taken to facilitate “human oversight” of AI procured by the government. LRI’s letters explain why summaries of AI models’ reasoning will not suffice for this purpose; the government needs access to models’ chains of thought and, in some cases, readouts of internal activations. Second, the clause requires contractors to report certain incidents to the government. LRI’s letters explain why the government needs to ensure that this requirement covers incidents initiated by models themselves during training or evaluation.
-
Read LRI’s August 2026 letter to GSA, as well as our April 2026 letter to GSA on an earlier version of the clause.
Personnel Vetting and AI Development
The White House has pledged to assist AI developers with securing their most advanced models. Among the types of cooperation foreshadowed was providing assistance with personnel vetting. Assistance with personnel vetting can help AI developers resolve two requirements that are otherwise in tension. On the one hand, they need to recruit and retain the talented foreign nationals that will allow them to maintain the U.S. lead in AI development. On the other hand, giving non-U.S.-person employees access to certain frontier models may create “deemed export” risks. Personnel vetting can be a key component of a technology control plan designed to let these employees continue their work while complying with regulatory requirements.
-
Read our Just Security article: Vetting Foreign AI Talent: Security Without Exclusion