Anthropic CEO Dario Amodei published a 3,800-word essay last Saturday titled "We Must Pace the Frontier," calling on the AI industry to deliberately slow the rate of model capability gains to buy 1–2 years for safety evaluation and alignment research.

He argued that since summer 2026, AI's ability to help build the next generation of AI has evolved rapidly, and without restraint the pace of breakthroughs will outrun human understanding and control. He cited the OpenAI model breach of Hugging Face as a warning sign.

Anthropic committed first: independent third-party evaluators will get permanent, employee-level access — badges, laptops, internal tools — and may publish findings with limited redactions. METR will serve as the first evaluator. OpenAI's Sam Altman responded within hours saying "I agree with Dario," and xAI's Elon Musk also voiced support.