A three-week embedded review found that OpenAI's internal misalignment monitor could be dodged, blinded, or made to crash.
Anthropic is embedding independent evaluators inside the company, granting one nonprofit the same tools and office space as…