Supporting case
Britain and Canada renew cooperation between their AI evaluation institutions
What happened
On October 13, Britain and Canada renewed their AI-security partnership. The official account identifies cooperation on standards, sharing information about frontier capabilities and risks, and regular exchanges between their evaluation institutions. Staff exchanges and coordinated evaluations are areas to explore, not completed deployments. Sensitive information may travel through appropriate Five Eyes channels. UK primary account ↗; Canadian counterpart ↗.
The renewal builds on agreements from 2024 and 2025. The published account supplies no new budget, developer access obligation, enforceable publication entitlement or public challenge process. It therefore establishes organized cooperation, not an independent international regulator or evidence that shared testing has already constrained a model release.
For The Race, public institutions can combine scarce evaluation expertise across borders. Who gets access to sensitive findings, and who can act on them, remains a separate question from the existence of a partnership.
Dates and implementation stage
Development: 2026-10-13
