The three AI companies are working to create an organization in which builders of frontier AI models would regulate themselves, without government oversight. They initially aimed to form a public-private partnership, but that effort stalled in the Trump administration, according to the report.
The standards body would support third-party organizations that test models before they are deployed, would detail how AI developers should report safety and security incidents, would define the labs’ voluntary safety and security commitments, and would establish qualifications for independent auditors of models and labs, the report said.
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
Members of a working group focused on the standards body are also considering having the group itself test models’ safety and capabilities, per the report.
Anthropic, Google, OpenAI and Microsoft created a nonprofit organization with many of these goals, the Frontier Model Forum, in 2023, and that group remains active. That organization and the potential new one are in contact and could work together, according to the report.
Critics of the plan for a new organization are concerned that Anthropic, Google and OpenAI could use it to box out open-source model developers and other competitors, the report said.
Bloomberg reported Wednesday (Sept. 24) that Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman addressed a United Nations Security Council meeting focused on concerns about AI and agreed that there is a need for global standards to address potential risks around the technology.
Bloomberg also reported Wednesday that Microsoft President Brad Smith told Bloomberg Television that Microsoft supports the use of independent evaluators for AI and that the company favors an off switch that would allow humans to retain control over advanced systems.
The current discussions of AI safety and standards follow a series of reports of incidents in which models from Anthropic, Google, OpenAI and Meta accessed the internet and hacked other companies. In some cases, companies did not publicly disclose the incidents or did not fully disclose them.
Add as Preferred Source
