- Sam Altman used a Financial Times op-ed to call for a US-led forum that would set AI safety standards and decide who can use the most advanced models.
- The op-ed landed alongside an FT report that OpenAI discussed giving the US government a stake of about 5 percent.
- It follows a June of regulatory pressure, including the pulling of Anthropic‘s Mythos and Fable and a reining-in of GPT-5.6’s release.
- The proposal shifts from voluntary commitments toward a body with real authority over model access.
What Happened
OpenAI chief executive Sam Altman called for a US-led forum with authority to set AI safety standards and decide who can access the most advanced models, in a Financial Times op-ed reported by The Rundown on July 3, 2026. The op-ed arrived alongside a separate FT report that OpenAI had discussed giving the US government a stake of about 5 percent.
The pitch, in Altman’s framing, is to invite the government in officially — starting with rules and a forum that has real regulatory teeth, rather than the voluntary pledges that have characterized industry self-governance so far.
Why It Matters
The proposal marks a shift from the voluntary safety commitments labs offered in 2023 and 2024 toward a government-backed body that would gate access to frontier systems. That is a meaningful concession from the CEO of the company with the most to lose from external control, and it reframes the safety debate around who decides access rather than what each lab promises.
It also follows a June in which Washington pressed the labs directly: Anthropic’s Mythos and Fable models were pulled and the release of OpenAI’s GPT-5.6 was reined in. Against that backdrop, Altman’s op-ed reads less as an abstract policy proposal than as a response to pressure already being applied. An equity stake, if it materialized, would give the public a direct financial interest in OpenAI’s upside — and give the government leverage that goes well beyond writing rules.
Technical Details
Altman’s op-ed grew out of June’s G7-adjacent discussions, according to The Rundown’s account. Two distinct threads are in play and should not be conflated: a proposed US-led forum that would set safety standards and control who can use the most capable models, and a separately reported discussion of a roughly 5 percent government stake in OpenAI. The two are not formally linked, and neither has been enacted. The forum concept centers on access control — deciding which organizations may run the most advanced models — rather than on model-level technical standards alone, which distinguishes it from earlier proposals focused on evaluations and disclosure.
Who’s Affected
A forum with authority over model access would affect every frontier lab, including OpenAI, Anthropic, and Google DeepMind, as well as the enterprises and developers that build on their models and would inherit whatever access rules the body sets. A government equity stake would specifically affect OpenAI’s investors and cap table, and would raise questions about conflicts between a regulator and a shareholder housed in the same government. Regulators outside the US would also have to decide whether to align with or diverge from a US-led standard, with implications for how models are sold internationally.
What’s Next
Both proposals remain at the discussion stage: the op-ed is a public pitch, and the 5 percent stake was described as something OpenAI discussed rather than agreed. The immediate test is whether other labs and lawmakers endorse a US-led forum or push instead for an international body that does not concentrate authority in Washington. The pulling of Anthropic’s Mythos and Fable and the constrained GPT-5.6 rollout suggest regulators are already acting without waiting for a formal structure, which raises the stakes on what shape that structure eventually takes.
