SAN FRANCISCO —With credit unions as concerned as any organizations over what advances in AI could mean to the security of data systems, OpenAI, Google and Anthropic have been holding talks about creating an industry standards organization that could establish common rules for testing and auditing advanced artificial intelligence systems, according to numerous media reports, as concerns are raised by some that AI is reaching a point where it could be dangerous to humans.
Representatives of the three leading AI companies have been meeting regularly since July to discuss the potential standards body, The Information reported Sunday, citing people familiar with the discussions.
The talks began before Anthropic CEO Dario Amodei publicly called over the weekend for AI companies to cooperate on testing and auditing increasingly powerful AI systems, according to the report.
The discussions represent a potentially significant step toward self-regulation by an industry facing growing questions about whether the development of increasingly capable AI models is outpacing efforts to evaluate and control their risks.
Altman Backs Industry Testing Organization
OpenAI CEO Sam Altman recently told employees during a companywide town hall that he supports establishing an organization to test and audit AI systems, The Information reported.
Altman, however, said he believes the major AI laboratories would have to create such an organization themselves rather than rely on support from the U.S. government, according to a person familiar with his comments cited by The Information.
The discussions come as some of the industry’s most prominent executives are publicly calling for greater cooperation among companies developing frontier AI models.
Amodei published a proposal Saturday calling for a slowdown in AI development amid concerns about potential safety risks.
As part of that proposal, Amodei argued that AI companies should voluntarily establish common safety standards while governments continue developing regulations for the rapidly evolving technology.
The proposal drew support from several prominent technology executives, including Altman, Google DeepMind CEO Demis Hassabis and Elon Musk, according to The Information.
Talks Reportedly Underway Since July
The idea of creating an industry-led organization to oversee AI standards is not new.
Hassabis proposed a self-regulatory AI organization in July modeled in part on the Financial Industry Regulatory Authority, or FINRA, which oversees U.S. broker-dealers.
What had not previously been disclosed, according to The Information, was the extent to which major AI developers had already begun discussing such an organization.
Representatives from Google, Anthropic and OpenAI have been meeting regularly since July about establishing an industry standards body, people familiar with the talks told the publication.
Such an organization could potentially provide a mechanism for competing AI developers to agree on common testing, auditing and safety practices even as federal policymakers continue debating what role government should play in regulating frontier AI.
President Trump has indicated he does not want to see limits placed on the development of AI as the United States seeks to compete with China.
