Independent third-party AI evaluators are gaining importance as governments and technology companies seek ways to assess model safety, while questions remain about their impartiality when companies pay them to evaluate their own models. Anthropic and OpenAI have sought help from smaller groups, including Model Evaluation and Threat Research and Apollo Research, to assess whether their models meet safety standards.

The article says AI researchers and technology executives have raised growing concerns as AI technology evolves rapidly. Anthropic CEO Dario Amodei emphasized third-party evaluation in an essay about pacing the AI frontier, and OpenAI CEO Sam Altman also supported the role of independent evaluators. President Donald Trump and Nvidia CEO Jensen Huang have dismissed calls for greater regulation, while Trump has praised companies’ “self-policing” and indicated he would largely leave the industry to its own devices.

Trump also encouraged AI companies to partner with an independent external auditor or evaluator under a voluntary accord he presented in late September. Government-licensed independent verification organisations (IVOs) could be authorised to test whether companies meet specific safety criteria; the groups have a role in the Frontier Act, introduced in July. A CNBC tech report said lawmakers in California, Connecticut and Virginia have taken steps toward implementing IVOs, while Massachusetts is considering broader independent AI safety evaluations. OpenAI and Anthropic support IVOs, but competition between them and evaluators’ financial ties to the companies they assess raise questions about impartiality.

Independent third-party AI evaluators are gaining importance as governments and technology companies seek ways to assess model safety, while questions remain about their impartiality when companies pay them to evaluate their own models. Anthropic and OpenAI have sought help from smaller groups, including Model Evaluation and Threat Research and Apollo Research, to assess whether their models meet safety standards.

The article says AI researchers and technology executives have raised growing concerns as AI technology evolves rapidly. Anthropic CEO Dario Amodei emphasized third-party evaluation in an essay about pacing the AI frontier, and OpenAI CEO Sam Altman also supported the role of independent evaluators. President Donald Trump and Nvidia CEO Jensen Huang have dismissed calls for greater regulation, while Trump has praised companies’ “self-policing” and indicated he would largely leave the industry to its own devices.

Trump also encouraged AI companies to partner with an independent external auditor or evaluator under a voluntary accord he presented in late September. Government-licensed independent verification organisations (IVOs) could be authorised to test whether companies meet specific safety criteria; the groups have a role in the Frontier Act, introduced in July. A CNBC tech report said lawmakers in California, Connecticut and Virginia have taken steps toward implementing IVOs, while Massachusetts is considering broader independent AI safety evaluations. OpenAI and Anthropic support IVOs, but competition between them and evaluators’ financial ties to the companies they assess raise questions about impartiality.