OpenAI, Anthropic, and Google DeepMind have established a joint AI standards body to develop shared safety benchmarks and evaluation protocols ahead of federal regulation.
The three artificial intelligence developers formally announced the joint organization on Thursday, publishing common testing protocols for frontier models. The new framework arrives as major stakeholders face mounting regulatory pressure in Washington, with the Department of Justice examining competition and safety practices across the sector.
OpenAI, Anthropic, and Google DeepMind Private Working Groups
The formal launch follows weeks of private coordination between the labs. OpenAI global policy chief Chris Lehane confirmed during meetings in Washington that his company had been collaborating with Anthropic and Google DeepMind for several weeks, validating earlier reports of the safety talks. Lehane noted that the companies do not require a government waiver to pursue safety coordination, despite previous concerns raised regarding potential antitrust scrutiny.
Demis Hassabis initially proposed an industry standards body in July to screen advanced models and coordinate industry-wide slowdowns if safety risks escalated. Concurrently, an essay published by Anthropic CEO Dario Amodei called on the industry to slow frontier model development to mitigate catastrophic outcomes. OpenAI’s Sam Altman, Google’s Demis Hassabis, and Elon Musk all expressed support for Amodei’s proposal, with Altman indicating OpenAI would join Anthropic in embedding third-party evaluators to monitor safety internally.
Shared Testing Protocols and Founding Supporters
The newly established organization will oversee evaluations covering cybersecurity misuse, biological risk, and deceptive model behavior. While OpenAI, Anthropic, and Google DeepMind anchor the initiative, Microsoft and Amazon were named as founding supporters through their cloud infrastructure partnerships and major stakes in the sector. Participation in the standards body remains open to other developers who meet established membership criteria.
The initiative mirrors earlier self-regulatory efforts such as the Frontier Model Forum. However, analysis of current governance frameworks notes that all three primary labs land at C-plus or below on the Future of Life Institute’s 2026 AI Safety Index, raising questions regarding how a new body will enforce rigorous testing standards beyond existing mechanisms.
Regulatory Scrutiny and Washington Response
The timing of the announcement coincides with shifting oversight in Washington. Attorney General Todd Blanche stated that the DOJ would examine competition and safety practices among leading developers. Meanwhile, House Speaker Mike Johnson remarked that Congress will not lead on AI safety, and President Trump dismissed safety concerns by warning that slowing development would benefit China.
Civil-society organizations immediately criticized the industry-led approach. Consumer groups and the Center for AI Safety argued that voluntary frameworks written by the largest firms risk entrenching market positions rather than securing public protection. Lehane stated that OpenAI supports a provision in the FRONTIER Act requiring frontier labs to permit independent verification organizations to assess their safety practices, while Microsoft endorsed a voluntary pacing framework.
