Policy

Our Policy Work

The Canary Institute develops evidence-based policy recommendations that translate technical AI safety research into actionable guidance for legislators, regulators, and the public. We aim to provide the kind of analysis that is currently missing from policy debates: technically grounded, politically independent, and oriented toward the public interest.

Safety Standards & Evaluation

We advocate for mandatory, independent safety evaluations of frontier AI systems before deployment. Current industry practice relies heavily on voluntary commitments and self-assessment. History offers no examples of a transformative technology that was made safe through voluntary compliance alone. AI systems with the potential for catastrophic misuse should meet clear, enforceable safety standards—just as pharmaceuticals, aircraft, and nuclear reactors do.

International Coordination

AI development is a global endeavor, and safety standards must be coordinated internationally to prevent regulatory arbitrage and ensure comprehensive coverage. The concentration of advanced AI capabilities in a small number of organizations and nations creates both security risks and governance challenges that no single country can address alone. We support the development of international institutions and agreements modeled on precedents from nuclear non-proliferation and biosecurity.

Transparency & Accountability

We support mandatory disclosure requirements for high-risk AI systems, including documentation of training data, capability evaluations, known failure modes, and incident reporting. The public and its elected representatives cannot govern what they cannot see. Transparency is not a burden on innovation—it is a prerequisite for trust.

Economic Transition

The economic disruption from AI-driven automation may be rapid and widespread. Policy frameworks must be developed now—not after mass displacement has occurred—to ensure that the productivity gains from AI are broadly shared. This includes investment in workforce transition programs, updated social safety nets, and serious examination of how the gains from AI-driven productivity are distributed.

Position Papers

Our policy briefs and position papers will be published here as they are completed, with accompanying public-facing summaries on our blog. Subscribe for updates.