Anthropic spends $900,000 on California Democrats: AI no longer just lobbies, it also funds parties

By The San Francisco Standard · October 31, 2026.

Anthropic sent $900,000 to California Democratic committees in the final stretch of the campaign, according to a late-contribution disclosure form the company filed Friday with the Secretary of State, as reported by The San Francisco Standard. Most of it, $525,000, went to the California Democratic Party. Another $150,000 went to a get-out-the-vote group.

It also gave $45,000 each to the Democratic parties of San Francisco, San Mateo, Orange, Fresno and Siskiyou counties. According to the outlet, the state party says the money will go toward electing Democrats up and down the ballot, fighting Proposition 39 (backed by Trump) and maintaining its voter protection programs.

The most striking detail is how the money moved. San Francisco party chair Nancy Tung says the donation made it possible to support two priority state Senate races, those of Melissa Hurtado (Central Valley) and Catherine Blakespear (Southern California), both contested by Republicans. According to the records cited, the party had already spent $400,000 on those two races before receiving Anthropic's check. Tung says the company had no say in where the money went and that there were no conditions attached.

It is one piece of a year of heavy political spending. According to the Standard, Anthropic had already given $850,000 to the state party and more than $100,000 to legislative candidates. It also contributed $500,000 to the campaign for Proposition H, Mayor Daniel Lurie's transportation measure, and another $500,000 to Proposition 1. Nationally, it has given $40 million to Public First Action, a group that advocates regulating AI. Its employees donated $3.83 million to federal campaigns in the first half of the year, according to the outlet's own analysis.

For scale, OpenAI matched the $500,000 on Propositions H and 1, and Meta has put $20 million into its own California political committee. But those contributions mostly go to ballot measures. Anthropic's bet is more directly partisan.

**Our take.** Anthropic is the subject of this story and of our analysis, so it is worth saying plainly: what matters is not whether the money is legal, but what model of influence it entrenches. Frontier labs have gone from lobbying behind closed doors to funding the electoral machinery. That evolution was predictable: whoever builds technology the state wants to regulate has enormous incentives to shape the people who will regulate it.

There is an argument in its favor. Anthropic advocates stronger safeguards for powerful systems, and funding those who share that position is a transparent, documented way to compete in public debate against other industry players that resist additional restrictions. Given the evidence-based regulation we advocate, having voices calling for safety is better than not having them.

But there is a near-term risk that should not be sugarcoated. When a company's money props up a party and that party will legislate on its industry, the perception of capture is inevitable, even if, as Tung says, there are no conditions. Public trust in AI governance depends on regulation not being seen as tailor-made for one of the competitors. The fact that the money also flows to contested races far from San Francisco shows how counterintuitive it is to track.

In the long run, if AI is to bring abundance and health for all, we need institutions that govern it with legitimacy. That requires more transparency, not less: it is worth watching the full disclosures the counties must file in January, and checking whether the spending pattern changes after the election.