Anthropic ties to AI safety nonprofits raise concerns over independence

Anthropic has aligned with several nonprofit groups that focus on AI safety, including Redwood Research and Model Evaluation and Threat Research (METR). Redwood, based in Berkeley, received multiple grants from the Effective Altruism-linked fund Coefficient Giving, which was co-founded by Redwood board member Holden Karnofsky. METR, originally part of the Alignment Research Center, also receives funding from the same source.

Critics argue that the overlapping leadership and financial connections create potential conflicts of interest. Perry Metzger, chairman of the Alliance for the Future, said the arrangement "does not provide an independent arbitrator for the AI industry" and suggested that the groups involved are closely linked.

Redwood’s CEO Buck Shlegeris responded that the organization’s work with AI companies has primarily been collaborative research rather than external accountability, and that it follows METR’s conflict-of-interest policies when partnering on investigations.

Anthropic announced a partnership with Accenture to embed safety evaluators within the company and said it is in dialogue with METR and other nonprofit evaluators about pilot projects. The firm declined to confirm ongoing negotiations with Redwood.

Lawmakers have expressed skepticism about industry-led oversight. Rep. Josh Gottheimer, co-chair of the House Commission on AI, warned that rapid adoption of self-regulatory frameworks by AI developers could be a red flag. Rep. Steve Scalise questioned the suitability of the groups involved for overseeing AI development.

The debate highlights ongoing concerns about whether AI safety oversight can remain impartial when funding and personnel overlap among industry players and nonprofit watchdogs.