Anthropic, OpenAI staff outdonate prior Big Tech cohorts

Something worth chewing on from this week's San Francisco Standard analysis: employees at Anthropic and OpenAI are giving to federal campaigns at rates that comfortably beat the Google, Facebook, and Airbnb workforces at the equivalent point in their own post-IPO cycles. Roughly 59 of every 1,000 Anthropic employees have donated this cycle, nearly triple Airbnb's rate and five to six times Facebook or Google. Anthropic staff have already moved about $3.83M into federal races, and adjusted for inflation, the Standard reports they are outgiving prior tech-IPO donor classes by more than threefold.

That intensity is not evenly spread. Nearly four in ten Anthropic donors, 39%, have maxed out to at least one campaign. In a single day last October, 20 Anthropic staff and 8 OpenAI staff combined to route about $173,000 to Alex Bores, a New York congressional candidate who championed AI safety legislation. The Standard traces the coordination back to posts on LessWrong, the AI safety community's discussion forum, that named specific candidates to support.

Why this matters if you are not tracking primary races: this is hard money, meaning it flows directly to campaigns rather than sitting in a super PAC, and it is the currency that pays for staff, ads under the campaign's own control, and candidate loyalty. It sits alongside much softer, louder money. OpenAI President Greg Brockman, per the same Standard analysis, gave two mega-donations of $12.5M apiece, one to Leading the Future, an anti-regulatory AI super PAC, and one to MAGA Inc, the Trump-aligned super PAC. Even excluding those gifts, OpenAI employees have surpassed giving from Google and Airbnb employees at the same post-IPO stage.

The honest caveat is the usual one for cycle-in-progress reporting. The Standard's numbers are a snapshot, and the cross-cohort comparisons sit on top of workforce definitions and inflation adjustments a reader has to take on trust. What the reporting does not give you is how concentrated the giving is among a small circle of early employees with liquidity, or how much coordination is happening off LessWrong in private group chats. Take the specifics as reported, not settled.

The forward-looking part is who this empowers. Candidates running on AI safety platforms now have a geographically concentrated, repeatable donor list that simply did not exist in the last cycle, and a handful of well-funded primaries in blue districts could push the topic onto the House floor faster than any lobbying budget alone.