Anthropic's principles, on retainer

ANTHROPIC spent $1.6m on federal lobbying in the first quarter of 2026, 4.4 times what it spent in the same period a year earlier and, for the first time, more than OpenAI. Three years ago its policy work consisted largely of academic essays on AI risk.

The trajectory is steep. In all of 2025 the company spent $3.1m on lobbying; in Q1 2026 alone it has matched roughly half of that figure, according to disclosures compiled by Issue One, a bipartisan watchdog. Most coverage has read the surge as a coming-of-age — the safety lab finally learning to defend itself in a hostile administration. Six lobbying shops have been retained since November 2024. The most recent addition has been the most consequential.

Acceptable use

The trouble is that the coming-of-age reading takes Anthropic's safety work and the lobbying surge as separate categories. Read alongside the calendar — Pentagon designation, Ballard hire, Wiles meeting, Trump walkback, all inside seven weeks — the disclosures are the operational record of the safety strategy itself. The principles and the apparatus to defend them have always traveled together; Q1 2026 is the quarter the spending caught up to the stakes.