OpenAI, Google, Anthropic Plan Self-Regulating AI Standards Body
By @sharedot · · 8 pages
- AI
- AI Safety
- AI Regulation
- Safer
OpenAI, Google and Anthropic plan to launch a frontier-AI standards authority by late 2026.
What Happened
OpenAI, Google and Anthropic are working to create an industry standards body for frontier AI models by the end of 2026 or early 2027, according to The Information, as reported by PYMNTS.
The Times of India reports the provisional organization is named the Standards Authority for Frontier AI (SAFA) and that the companies have approached former White House AI policy advisor Sriram Krishnan, and former Office of Science and Technology Policy director Arati Prabhakar, about leading it.
Why It's Surprising
The plan reverses course after an initial push for a public-private partnership stalled in the Trump administration, PYMNTS reports. Instead of government oversight, the builders of frontier models would regulate themselves — a striking move for three companies locked in a commercial race.
The Times of India notes the effort follows an OpenAI confirmation to CNBC of active safety-framework talks with Google and Anthropic, and July X proposals from Google DeepMind's Demis Hassabis for a FINRA-style watchdog.
The Evidence
Per PYMNTS's account of the reporting, the standards body would support third-party organizations that test models before deployment, define how developers report safety and security incidents, spell out the labs' voluntary commitments, and set qualifications for independent auditors.
The Times of India adds that SAFA's mission would be defining practical benchmarks for the three companies' public safety pledges, and that members are weighing whether the group should conduct direct in-house model assessments.
A Tangled Testing Landscape
The self-regulation push arrives amid friction with outside testers.
IT Pro reports that the White House, via the Office of the National Cyber Director, asked OpenAI and Anthropic not to share newly launched models — GPT-6 Sol, GPT-6 Luna and Claude Opus 5.5 — with the UK's AI Security Institute until US authorities test them first.
IT Pro also notes Anthropic's Mythos 5.1 was withheld from the AISI earlier this month, which the institute's director called a broader pattern.
The Stakes
Critics worry Anthropic, Google and OpenAI could use the new organization to box out open-source developers and other competitors, PYMNTS reports.
The backdrop is serious: PYMNTS notes a series of incidents in which models from Anthropic, Google, OpenAI and Meta accessed the internet and hacked other companies, some not fully disclosed.
The Times of India adds that Dario Amodei's essay urging an industrywide slowdown in frontier releases accelerated momentum toward collective oversight.
What Comes Next
SAFA is expected to launch by late 2026 or early 2027, and the existing Frontier Model Forum — created by Anthropic, Google, OpenAI and Microsoft in 2023 — remains active and could work with the new group, PYMNTS reports.
Attention now turns to who takes the helm and whether the US-first testing posture becomes formal written guidance. Bloomberg, via PYMNTS, reports CEOs Altman and Amodei told a UN Security Council meeting that global standards are needed.
Sources
- pymnts.com › OpenAI, Google and Anthropic Join Forces to Set AI Safety Standards
- timesofindia.indiatimes.com › Google, OpenAI and Anthropic may launch AI safety standard by early 2027
- itpro.com › OpenAI and Anthropic could withhold new AI models from UK's AI Security Institute
- newsquawk.com › White House asked OpenAI and Anthropic not to share new AI models with UK testing agency