US officials have held preliminary talks about the possibility of the government taking equity stakes in artificial intelligence companies OpenAI and Anthropic, according to people familiar with the discussions. The conversations remain at an early stage, and no formal proposal has been presented to either company.
What the talks cover
The discussions center on whether the federal government could acquire ownership shares in the two leading AI labs as part of a broader effort to keep American AI development ahead of foreign rivals, particularly China. One idea under consideration is that a government stake could give Washington a seat at the table — or at least formal oversight — over how the most advanced AI models are built and deployed.
Neither OpenAI nor Anthropic has commented publicly on the matter. Both companies have been vocal about the need for responsible AI governance, but have also pushed back against what they see as overly prescriptive regulation.
Why these companies
OpenAI and Anthropic are two of the most prominent players in the race to build general-purpose AI systems. OpenAI, backed by Microsoft, created ChatGPT and the GPT-4 model. Anthropic, founded by former OpenAI employees, has developed its own line of large language models called Claude. Their technologies are seen as critical to both economic competitiveness and national security.
Government equity stakes would represent a significant departure from the usual relationship between Washington and Silicon Valley. Historically, the US has funded basic research and bought products, but rarely taken direct ownership in private technology companies. The current discussions reflect a growing anxiety among policymakers about ceding leadership in AI to China.
How a stake might work
Officials are exploring multiple structures, according to the people familiar. A direct purchase of shares is one option, though that would require congressional approval and a specific appropriation. Another path might involve the government receiving warrants or convertible notes tied to future funding rounds.
Taxpayer investment would come with strings attached. Any equity deal would likely include governance rights, data access for safety testing, and restrictions on selling the technology to adversaries. The details are far from settled, and the legal framework for such an arrangement does not currently exist.
What happens next
The talks are happening inside the White House and at several federal agencies, but they have not yet reached a stage where either company has been formally approached with a written proposal. That could change as Congress returns to session and begins drafting AI legislation this fall.
A key open question is whether lawmakers on both sides will support putting taxpayers' money into private AI companies — and under what conditions. For now, the discussions remain just that: discussions, with no deadline and no decision expected in the immediate term.
