OpenAI, Google and Anthropic are working together to launch an independent standards body for artificial-intelligence safety, The Information reported on Thursday, citing unnamed sources.
The proposed organisation would be a self-regulatory body run by the builders of frontier AI models without government oversight. It would support independent third-party testing of models before they are deployed, define how AI developers should report safety and security incidents, set out the labs' voluntary safety and security commitments, and establish qualifications for independent auditors of models and labs. Members of the working group are also considering whether the body itself should test models' safety and capabilities.
## Stalled government route
According to the report, the three companies initially aimed to form a public-private partnership, but that effort stalled inside the Trump administration. The new initiative is designed to go further than the Frontier Model Forum — the nonprofit the companies co-founded with Microsoft in 2023 to advance AI safety research — creating something with more structure, independence and teeth, though the Forum remains active and the two efforts are said to be in contact.
The talks are being handled by teams below chief-executive level and have been underway since July, according to coverage of the report.
## UN Security Council link
Bloomberg reported that Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman separately addressed a United Nations Security Council meeting focused on AI risks last week, with both agreeing there is a need for global standards to address potential risks around the technology.
## Credibility questions
Coverage of the plan also flagged tensions. The Future of Life Institute's 2026 AI Safety Index scored all three companies C+ or lower on transparency, safety implementation and responsible development practices. No formal organisation, leadership, funding structure or charter exists yet, and critics worry the standards could be written to suit giant labs while boxing out open-source model developers and smaller competitors.
The companies have not confirmed the plan, and the proposed launch — targeted for the end of this year or early next year — depends on the working group resolving differences over governance, funding and structure.
Sources: [PYMNTS report](https://www.pymnts.com/news/artificial-intelligence/2026/openai-google-and-anthropic-join-forces-to-set-ai-safety-standards/) (PYMNTS, September 24); [CryptoBriefing report](https://cryptobriefing.com/google-openai-anthropic-ai-safety-standards/) (CryptoBriefing)
The proposed organisation would be a self-regulatory body run by the builders of frontier AI models without government oversight. It would support independent third-party testing of models before they are deployed, define how AI developers should report safety and security incidents, set out the labs' voluntary safety and security commitments, and establish qualifications for independent auditors of models and labs. Members of the working group are also considering whether the body itself should test models' safety and capabilities.
## Stalled government route
According to the report, the three companies initially aimed to form a public-private partnership, but that effort stalled inside the Trump administration. The new initiative is designed to go further than the Frontier Model Forum — the nonprofit the companies co-founded with Microsoft in 2023 to advance AI safety research — creating something with more structure, independence and teeth, though the Forum remains active and the two efforts are said to be in contact.
The talks are being handled by teams below chief-executive level and have been underway since July, according to coverage of the report.
## UN Security Council link
Bloomberg reported that Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman separately addressed a United Nations Security Council meeting focused on AI risks last week, with both agreeing there is a need for global standards to address potential risks around the technology.
## Credibility questions
Coverage of the plan also flagged tensions. The Future of Life Institute's 2026 AI Safety Index scored all three companies C+ or lower on transparency, safety implementation and responsible development practices. No formal organisation, leadership, funding structure or charter exists yet, and critics worry the standards could be written to suit giant labs while boxing out open-source model developers and smaller competitors.
The companies have not confirmed the plan, and the proposed launch — targeted for the end of this year or early next year — depends on the working group resolving differences over governance, funding and structure.
Sources: [PYMNTS report](https://www.pymnts.com/news/artificial-intelligence/2026/openai-google-and-anthropic-join-forces-to-set-ai-safety-standards/) (PYMNTS, September 24); [CryptoBriefing report](https://cryptobriefing.com/google-openai-anthropic-ai-safety-standards/) (CryptoBriefing)