Altman Wants Washington As AI Shareholder.

Mazen Al Sabbagh

— 4 min read
Share

TL;DR

  • OpenAI Equity Bid: Sam Altman proposes a 5% US government stake in leading American AI labs, including OpenAI, alongside advocating for a Washington-led international AI governance forum.
  • Meta's Agent Lag: Leaked town hall audio reveals Mark Zuckerberg's disappointment with AI agent progress, despite claims that an unreleased "Watermelon" model matches GPT-5.5 benchmarks. Meta shares declined 4.9%.
  • Policy Alignment: The White House is nearing an agreement with OpenAI, Google, and Anthropic on voluntary pre-release security reviews for frontier AI models, with an announcement expected next week.
  • Anthropic Expansion: Following the lifting of export controls, Claude Fable 5 is now globally available. Claude Sonnet 5 also launched as the new default model, with introductory pricing.
  • Market Dynamics: June payrolls significantly missed consensus, and a tech-led selloff, particularly in semiconductors, marked a cautious start to the second half after an 80% first-half run.

Lead Story: Altman Wants Washington As AI Shareholder.

Sam Altman has positioned the US government as both potential regulator and equity partner in the AI sector. His Financial Times op-ed advocates for a US-led international forum to establish standards and offer impartial risk analysis, citing precedents like aviation safety bodies and the IAEA. This suggests a framework for controlled global AI deployment.

This proposal gains tangible form with OpenAI's preliminary discussions to grant the US government a 5% equity stake. Valued at approximately $42.6 billion based on its $852 billion valuation, this model, akin to the Alaska Permanent Fund, has been pitched to key administration officials. Altman intends for this 5% government ownership to extend to all leading US AI developers, though other companies have not committed.

The strategic timing is clear. Fortune's analysis highlights Anthropic's reported revenue run-rate surpassing OpenAI's, alongside a dip in ChatGPT's generative-AI web visit dominance. With an S-1 filed and a September IPO targeted, this statesmanship represents a calculated maneuver, providing regulatory de-risking ahead of a public listing.

In Other News

Zuckerberg admits Meta's agent bet has stalled — as Wang touts "Watermelon." Leaked audio from Meta's internal town hall reveals Mark Zuckerberg's assessment that AI agent development "has not accelerated as anticipated" over the last four months, following an 8,000-person reorganization less than two months prior. Despite this, and a projected $125–145 billion in 2026 AI capex, he expects returns in three to six months. Concurrently, Superintelligence Chief Alexandr Wang claimed their forthcoming "Watermelon" model has matched GPT-5.5 on unspecified benchmarks. Meta shares concluded down 4.9% at $582.90.

Fable 5 goes global as Commerce lifts export controls. The Commerce Department has removed export controls on Anthropic's Claude Fable 5 and Mythos 5, ending a brief restriction. This followed a jailbreak discovery by Amazon researchers that allowed Fable 5 to generate exploit code. In return, Anthropic committed to proactive security detection, threat intelligence sharing, and co-developing standards with the government. The global rollout commenced Wednesday, shortly after Claude Sonnet 5 launched as the default model, priced at $2/$10 per million tokens until August.

White House nears voluntary frontier-model standards. The White House is in advanced discussions with OpenAI, Google, and Anthropic to establish a voluntary framework for pre-release security reviews of frontier AI models. New models would undergo up to 30 days of government scrutiny before deployment, with an announcement potentially arriving the week of July 7. While optional, non-participation could impact federal contracts. Google is currently negotiating ahead of its Gemini 3.5 Pro release, delayed from June due to enterprise concerns over token consumption in agentic tasks.

Jobs miss caps a wobbly start to the half. June saw the addition of only 57,000 jobs, significantly below consensus, with May's figures also revised downwards. Unemployment declined to 4.2%, primarily due to a drop in participation to 61.5%. This prompted traders to adjust September rate-hike expectations. The jobs report followed a tech-led market selloff on Wednesday, where the tech sector SPDR fell 2.6%, and semiconductor firms like SanDisk and Applied Materials saw significant declines as investors rotated out of positions after tech's 80% first-half run.

X / Social Pulse

Zuckerberg’s leaked comments dominated X, juxtaposing his prior "most consequential technology" memo with today’s "hasn't really accelerated" remark. IBTimes' "The $145 Billion Lie?" narrative garnered significant traction, as did skepticism regarding Meta's "Watermelon" claims without disclosed benchmarks. Altman's 5% equity proposal drew mixed reactions, oscillating between calls for public upside and warnings of regulatory capture, with VCs noting the op-ed and equity story's coincident timing.

One to Watch

GPT-5.6. OpenAI previewed GPT-5.6 on June 26, powered by Cerebras hardware, boasting 750 tokens per second. Deployment to select clients is slated for this month. With Anthropic's Sonnet 5 already available, Gemini 3.5 Pro imminent, and Meta's "Watermelon" claims circulating, OpenAI faces a narrowing window for its next flagship. The September IPO timeline significantly amplifies public scrutiny.

Quick Hits

The through-line today is equity — literal, in Altman's 5% offer, and figurative, in who gets to ship frontier models on what terms. Between the stake talks, the export-control reversal, and the voluntary-standards framework, Washington is becoming a counterparty to the AI race, not just its referee.

Sources

OpenAI / Altman: Fortune · Bloomberg · CNBC · Forbes · Techmeme Meta: Implicator · IBTimes UK · Yahoo Finance Anthropic: CNBC · Al Jazeera · The Hacker News · Anthropic · VentureBeat Policy: Yahoo Finance / FT · TNW Markets & jobs: CNBC jobs · NBC News · Yahoo Finance · CNBC H1 recap Quick hits: The Register · SecurityWeek · Build Fast with AI · TechCrunch · Gov.ca.gov

Lock in. M. mazen@thorterminal.com

Read more