Anthropic CEO Dario Amodei announced the company will provide third-party evaluators with permanent, employee-level system access. This initiative allows outside experts to verify adherence to safety protocols throughout the development process.
The commitment marks the first phase of Amodei’s "We Must Pace the Frontier" proposal. This plan advocates for a deliberate slowdown in AI advancement to prioritize safety.
The new process will assess risks within both finished models and underlying training pipelines. This transparency strategy places competitive pressure on rivals Google and Microsoft-backed OpenAI to adopt similar safety standards.