A coalition of researchers calls on AI developers to transparently share safety methods, evaluations, and data with the global scientific community to enable independent scrutiny and advance AI safety research. Transparency should include safety-relevant details about frontier models while allowing exceptions for genuine security risks.