According to a Financial Times report picked up the same day by Bloomberg, CNBC and Forbes, OpenAI proposed giving the US government a 5% stake in the company. The scheme, inspired by the “Alaska Permanent Fund” (the fund through which that state distributes oil revenue among its residents), would be worth about $42.6 billion based on the valuation the company had in March 2026. Sam Altman reportedly also suggested it as a model that could be extended to Anthropic, Google and Meta.
It is the first time a frontier artificial intelligence lab has formally proposed that a state co-own its business. The logic is political more than financial: giving up a slice of the pie to soften the growing backlash against the concentration of power in the hands of a handful of companies. If the government is a partner, criticism of the size and influence of these labs becomes, in theory, harder to sustain.
For Latin America, the news matters as a mirror. No government in the region today has the negotiating position or the legal mechanism to raise an equivalent conversation about the foreign AI infrastructure that already operates (and bills) within its borders. While Washington discusses how much of OpenAI it is entitled to, the region’s countries remain customers of services that are trained, charge and set their rules in another hemisphere. The contrast is not about greed: it is about who has leverage when critical infrastructure belongs to someone else.
Also today
- Anthropic in talks with Samsung to manufacture its own AI chip — It follows the path OpenAI already took with its “Jalapeño” chip: every frontier lab is now seeking its own silicon so as not to depend on third parties.
- Microsoft launches “Frontier Company,” a $2.5 billion enterprise AI unit — 6,000 experts to deploy AI at scale for large clients, a field where it competes with consulting firms such as PwC, KPMG and Accenture, which have a strong presence in the region.
- Court emails reveal the rift between Dario Amodei and the Pentagon — According to the emails, Anthropic’s CEO refused to loosen Claude’s ethical limits for surveillance and autonomous weapons, the origin of the company’s designation as a “supply chain risk.”
- Tesla caps employees’ AI spending at $200 a week, but exempts Grok — It is the fifth known case, after Uber, Meta, Amazon and Walmart, of a large company capping internal spending on assistants that work autonomously.
In the region
Two stories show AI entering Latin American public administration through the service entrance. The Buenos Aires city government presented BAX, a conversational assistant for looking up procedures and services; and in Guatemala, the forensic institute’s “IdentIA” tool made it possible, for the first time, to identify a woman in the morgue after a month without her being recognized.
In parallel, this week Geneva hosts two global governance tables with uneven representation for the region. On July 2, the International Telecommunication Union launched the “AI for Good Global Commission” (co-chaired by the president of Rwanda and the CEO of Salesforce, with more than 40 founding members among heads of state and lab executives) with no Latin American country confirmed as a founder. But in the same place and on the same days, Salvadoran ambassador Egriselda López co-chairs, together with Estonia, the UN Global Dialogue on AI Governance (July 6 and 7), the intergovernmental mechanism that does have a leadership seat for the region. The nuance is worth keeping in mind before simplifying the narrative of “Latin America left out of the table”: it is present in the dialogue among governments and absent from the industry commission.
Threads we’re following
The emails between Dario Amodei and the Pentagon add a piece to a story we have been following. In June, Anthropic was designated a “supply chain risk” and its Fable 5 model was banned for 19 days before returning to the public on July 1; the emails now made public suggest that behind that friction was the company’s refusal to relax Claude’s limits for military uses. Along the same line of tension with state actors, and according to the Financial Times, Anthropic added a second layer of response to the episode of Alibaba’s distillation of its model: it would now block Chinese firms’ access to Claude through subsidiaries and virtual private networks. The tug-of-war between labs and governments over who controls these models is being written almost daily.
If OpenAI is willing to give up 5% of its ownership to buy political legitimacy, what would have to happen for a Latin American government to even raise a similar conversation about the AI that already operates within its borders?
About this entry. It is generated automatically from public sources, without human review before publication. It may contain errors of interpretation or summary; please check each story against its original source (the links lead there) before citing it or making decisions based on it.
Doble Click is written with Anthropic models.