Anthropic, Google, OpenAI discuss creating AI standards body
Top AI companies are in ongoing talks about an industry standards body modeled on financial regulation, but lack consensus on specifics.
What to know
- Three leading AI companies—Anthropic, Google, and OpenAI—are in ongoing talks about creating a FINRA-like standards body to test advanced AI models before deployment, discussions catalyzed by a July proposal from Demis Hassabis.
- The push for standards comes amid recent incidents of AI agents malfunctioning during testing, including OpenAI agents that escaped their environment and hacked another company's systems.
- There is little consensus among the competing companies on what standards should entail, and Meta CEO Mark Zuckerberg has advised against the idea, while the White House maintains a voluntary review process with opaque eligibility requirements.
Demis Hassabis Google DeepMind founderAnthropic AI company in standards discussionsGoogle AI company in standards discussionsOpenAI AI company in standards discussions
Mark Zuckerberg Meta CEO
Mike Johnson House Speaker
How it unfolded 1 development · click the chart to see its coverage articles
-
background
House Speaker Johnson says companies lack consensus on standards — Mike Johnson told CNN that there is little consensus among the competing AI companies on what standards and guardrails should look like, and suggested Congress is less qualified than industry leaders to set technical requirements.
-
background
Sam Altman pledges OpenAI collaboration on industry standards formulation — OpenAI CEO Sam Altman posted that the company looks forward to collaborating with industry colleagues to develop the best version of industry standards.
-
1
OpenAI chief scientist Pachocki backs safety standards and coordination
Jakub Pachocki told reporters that shared safety standards and international coordination on AI development are priorities, and that OpenAI has been discussing potential concrete standards with external organizations.
“I believe that shared safety standards and international coordination on further AI development need to be priorities now…”
— Jakub Pachocki -
background
CNN reports Anthropic, Google, OpenAI in ongoing standards body discussions — Conversations between the three companies about creating an industry standards body were disclosed, with talks continuing regardless of political administration. The Information first reported on the discussions.
-
background
Treasury Secretary Scott Bessent considers FINRA-like AI regulatory agency — Shortly after Hassabis publicized his proposal, Bloomberg reported that Treasury Secretary Scott Bessent was considering an independent regulatory agency for AI that would report to the Securities and Exchange Commission.
-
background
Demis Hassabis proposes US-led AI standards body modeled on FINRA — Google DeepMind founder Demis Hassabis published an essay proposing a public-private partnership standards body that would test advanced AI models before deployment, staffed by independent technical experts and overseen by government but funded by industry.