A viral analysis by researcher Kevin Bass, published on September 14–15, 2026, has prompted calls for a Congressional investigation into potential conflicts of interest within the AI safety sector. The report, which garnered nearly 5 million views, highlights the close ties between AI laboratories and the organizations responsible for evaluating their safety.
The controversy centers on METR (Model Evaluation and Threat Research), an organization that assesses frontier AI models. Critics point to a revolving door of staff between METR and companies like OpenAI and Anthropic. Additionally, funding for safety watchdogs, including METR and Redwood Research, is often linked to investors like Facebook co-founder Dustin Moskovitz, who also holds equity in Anthropic.
METR has stated that it does not accept donations or compensation from AI labs or their employees and claims to reject certain funding sources to preserve its independence. Supporters of the current model argue that the specialized nature of AI safety research limits the available talent pool, making it inevitable that experts have prior experience with major labs.
On September 18, 2026, more than 100 AI experts, including Geoffrey Hinton, signed an open letter demanding structural reforms. The signatories are calling for full editorial autonomy for evaluators, complete transparency regarding methodologies, and protections against retaliation from the companies they review. No regulatory changes have been implemented to date.
