Breaking

OpenAI Proposes Federal AI Rules Requiring Mandatory Reviews of Frontier Models

July 2, 2026 at 12:22 EDT

Policy & Regulation
Foundation Models

OpenAI's Sam Altman has proposed a federal framework centered on government approval and safety checks for the most capable AI systems, arguing in an FT op-ed and a series of policy documents that "determining safety standards is a prerequisite to broad distribution" of AI. Yet the plan is drawing criticism for presuming a US-led order.

AI Governance · Sam Altman, OpenAI

A US-led international forum to make advanced AI safe — and shared

Altman proposes that democratic institutions set the rules while labs build the technology — with access granted only to countries and companies that follow agreed safety standards.

1–2 yrs

Until systems of "astonishing power" arrive, Altman predicts

~5%

US government stake reportedly in preliminary talks

Precedents cited: aviation, finance, the IAEA

How the proposed forum would work

Set standards

Agreed rules & impartial risk analysis

→

Certify members

States & companies pass regular checks

→

Grant access

Technology shared with compliant parties

"Without shared safety standards, broad distribution of the technology will actually be held back."

A division of labour: democratic institutions make the rules, labs build the models — reconciling safety and access through cooperation rather than country-by-country restrictions.

Welcomed

Framing AI's wealth and benefits as something to be broadly shared — and hopes that common standards could accelerate adoption.

Criticised

Reads as a "permission regime" that could entrench governments and incumbents — compliance language dressed as public interest.

The proposal remains conceptual — no model names, participating countries, or certification procedures have been spelled out.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.