Our research, policy frameworks, and briefing materials — free to read, cite, and share.
A bipartisan position paper and legislative framework proposing the American Artificial Intelligence Leadership and Accountability Act, which would establish the Artificial Intelligence Oversight Commission. Includes a full section-by-section architecture, the bipartisan case, anticipated objections, and an implementation timeline.
A condensed summary of the proposed Commission and the Act for policymakers, staff, and press: the problem, the SEC precedent, the tier structure, and what the framework deliberately does not do.
A briefing paper documenting the current harms of large language models, the physical and social costs of the data center buildout, and the future risk of unaligned AI — with sources throughout. Written for donors, journalists, and anyone new to the field.
A plain-language presentation making the case for the AI Oversight Commission, built for general audiences: what AI is already doing today, why voluntary self-regulation isn't oversight, what the Act would actually require, and — just as importantly — what it deliberately leaves alone.
A three-minute explanation of the current harms, the regulatory gap, and the proposed Artificial Intelligence Oversight Commission.
A comprehensive, citation-backed reference on the risks of misaligned artificial general intelligence — capability timelines, failure scenarios, proposed solutions, and a full research glossary.
Visit AGIRisk.comOur own primer on the three interconnected research domains we fund — alignment, safety, and ethics — and why neglected research directions matter most.
Read the primerA plain-language walkthrough of the proposed AI Oversight Commission: the case for action, the 1934 precedent, the tier structure, and the framework's designed limits.
Read the summaryEvery document on this page was produced without corporate sponsorship. Your support keeps it that way.