Anthropic Selects Accenture as First Embedded AI Evaluator
What happened: Anthropic announced Accenture as its first embedded evaluator to red-team models and test AI safeguards, with both firms committing at least $1B over five years.
What happened: Anthropic announced Accenture as its first embedded evaluator to red-team models and test AI safeguards, with both firms committing at least $1B over five years. The partnership is non-exclusive, with additional evaluators expected. Anthropic will fund Accenture's work initially, with plans to shift to pooled or government funding.
Why it matters: The move operationalizes Dario Amodei's "AI slowdown" proposal, aiming to address risks of recursive self-improvement. However, the choice of a management consultancy over a dedicated AI-safety nonprofit has drawn skepticism about independence and rigor. Industry leaders remain divided on the need for such regulation.
Source: TechCrunch