Growing scrutiny of the revolving door between artificial intelligence safety evaluation organizations and the companies they are meant to oversee has brought conflict-of-interest concerns to the forefront. The people assessing AI safety may, in some cases, have professional or financial links to the very labs they evaluate. Because those assessments are meant to determine whether frontier models are safe enough for deployment, questions about evaluator independence now extend across the entire industry.

Proposals from Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman to embed third-party evaluators directly inside their laboratories have intensified the debate, turning a simmering concern into a prominent public issue. The structure would place watchdogs inside the very environments they are meant to scrutinize, sharpening questions about how independence would be preserved.

The web of connections

At the center of the controversy sits METR, or Model Evaluation and Threat Research, one of the most prominent organizations tasked with assessing whether frontier AI models are safe enough for deployment. METR's staff includes people who previously worked at OpenAI and Anthropic, and the overlap extends beyond individual resumes.

The funding picture is especially tangled. Dustin Moskovitz, the Facebook co-founder, holds equity in Anthropic. His philanthropic foundations channel money through Effective Altruism networks that have provided significant funding to evaluator organizations such as METR and Redwood Research. The result is an ecosystem in which money flowing to safety watchdogs can be traced back, sometimes through only a hop or two, to the companies being evaluated.

Several evaluators also have personal or familial ties to employees at the labs they assess.

Researcher Kevin Bass published an analysis on September 14–15, 2026, mapping these in detail. The thread attracted nearly 5 million views and included an explicit call for a Congressional investigation into what Bass characterized as structural conflicts of interest.

The industry pushes back

METR has responded directly to the allegations, stating it does not accept compensation or donations from AI labs or their staff. The organization said it has actively rejected certain funding sources to maintain its independence and emphasized its commitment to disclosing potential conflicts.

Defenders of the current arrangement note that evaluating frontier AI models requires extremely specialized technical knowledge. The talent pool is inherently narrow, and the people with the relevant skills have almost certainly worked at or been funded by AI labs at some point. That scarcity helps explain why the personnel and funding overlaps persist even among organizations committed to independence: the same small community builds frontier systems, staffs the evaluators, and sits upstream of the philanthropic money that funds them.

The expert letter and what comes next

On September 18, 2026, more than 100 AI experts signed an open letter demanding structural reforms. The signatories included Geoffrey Hinton, the Turing Award winner who has become one of the most vocal critics of how the AI industry self-polices.

The letter called for evaluators to have full editorial autonomy over their findings, complete transparency about methodologies and results, and explicit protection from retaliation by the companies they review.

No significant regulatory changes have been enacted so far, but the political pressure continues to build. The questions to watch include whether Congress acts on the investigation call raised in Bass's analysis, whether labs or evaluators move to adopt the letter's specific demands, and whether the embedded-evaluator proposals from Amodei and Altman take shape in any form.