# Anthropic Adds External AI Evaluators

Anthropic has brought six external evaluators from the nonprofit Model Evaluation and Threat Research (METR) into its AI assessment process. The article characterizes this as a step toward independent evaluation but argues that company selection of evaluators and a narrow emphasis on safety limit public accountability. It advocates publicly funded, democratically selected experts embedded in frontier AI development teams, with assessment covering social, psychological, workforce, surveillance, and democratic risks. No government action, procurement opportunity, or contract is announced.

- For agencies evaluating frontier AI systems, the debate highlights evaluator independence, selection processes, and the breadth of risks assessed as potential considerations in AI assurance procurements.
- The article’s proposal for publicly funded evaluators is an advocacy position, not an announced funding program or requirement; contractors should not treat it as a current solicitation or compliance mandate.
- AI assessment firms and prospective government buyers may find the discussion relevant when defining evaluation scope and accountability expectations, but the signal provides no specific procurement timeline or action.

**Jurisdictions:** federal
**Industries:** Information Technology
**Topics:** Artificial Intelligence
**Published:** October 23, 2026

### Key Quotes
> keep your eyes open about the direction things are heading and not necessarily leave it all up to the people that are building it because, you know, they want engagement.
> — Jacob Coxon, AI researcher

### Sources
- [Anthropic’s AI oversight isn’t democratic enough](https://thehill.com/opinion/technology/6165747-anthropic-external-evaluators-ai) - The Hill