Anthropic and Accenture to each invest $1B+ in embedded AI evaluators with near-employee access
eyishazyer · x · 2026-09-20
Anthropic is partnering with Faculty, Accenture's AI unit, to build an "embedded evaluation" program: external evaluators will get near-employee-level access to red-team models, run alignment assessments, and test safeguards while models are still in training.
- Both companies commit at least $1B each over five years to build this capacity.
- It's the first concrete step from Dario Amodei's "We Must Pace the Frontier" essay; Anthropic says it is separately in talks with METR.
- Anthropic acknowledges no standards yet exist for evaluator access or reporting, and stresses that model safety remains its own responsibility—embedded evaluation only makes accountability more verifiable.
More from Companies & People
- Developers frustrated as App Store won't feature dedicated iPhone Duo section — jdluk87 · 2026-09-20
- How big an account do you need before OpenAI and Anthropic offer early access? — AIandDesign · 2026-09-20
- Lawsuit alleges Anthropic, OpenAI and others made illegal agreement to slow down AI — embedding-shape · 2026-09-20
- Slack + Claude Suddenly Became Salesforce's Core 'We Serve AI Natives' Pitch — thedealdirector · 2026-09-20
- PE Firm Spent Six Figures on AI Consulting for 7 Portfolio Companies, Got a Chat Subscription — alex_verem · 2026-09-20
- 58 law firms bought AI, only one can tie spend to a matter — jkubicki · 2026-09-20