Neal & Leroy Advises METR on Launch of AI Risk Assessment Pilot and Inaugural Frontier Risk Report

Neal & Leroy advised Model Evaluation and Threat Research, Inc. (METR), a nonprofit organization focused on AI safety research, in connection with the structuring, agreement negotiation and launch of its groundbreaking industry risk assessment and the launch of its first Frontier Risk Report. Under the pilot, METR conducted independent risk assessments evaluating loss-of-control risks associated with frontier AI models developed by Anthropic, Google DeepMind, Meta, and OpenAI. The assessments involved technical access to each participant's models, review of non-public information regarding capabilities, alignment, and control measures, and the preparation of company-specific private reports and an industry-wide public report summarizing METR's findings across the participants.

The resulting Frontier Risk Report, spanning over 300 pages, represents the first independent, cross-industry assessment of loss-of-control risks from advanced AI agents and is intended to create an evidence base for use by industry, civil society, academia, and government in evaluating and responding to such risks. Read the report here.

The Neal & Leroy team was led by Matthew Sumner and MacKinzie Neal.