Anthropic and OpenAI set lobbying records in Q2 2026 federal filings

The fiercest AI competition this quarter was not on any benchmark. It was in Washington.

The numbers in the filings

Fresh federal filings show Anthropic spent $1.97 million on lobbying in Q2 2026, up 26 percent from the previous quarter. That is more than chip giant Nvidia at $1.25 million, and close to Oracle’s $2 million. OpenAI added $1.2 million of its own, up 18 percent.

Combined, the two labs hit $3.17 million for the quarter, roughly double what they spent a year ago. And the trajectory is steeper than any single quarter shows: Anthropic’s first half of 2026 alone, over $3.5 million, already beats everything it spent in all of 2025.

What the money is for

The filings list four battlegrounds: cybersecurity, copyright, cloud computing and defense procurement.

Every one of those is an existential question for AI companies right now. Copyright covers the training-data lawsuits that could reprice the entire industry. Defense procurement covers the military contracts becoming a major revenue line. Cybersecurity and cloud computing decide how, and on whose infrastructure, frontier models get deployed.

Why now

The timing is not subtle. Midterm elections are ahead, IPOs are on the horizon, and AI regulation is still being written. The rules that will govern the industry for a decade are being drafted in this window, and the labs are spending accordingly.

There is a cultural marker buried in the spreadsheets too: the labs that used to publish papers now hire lobbyists. An industry that once treated Washington as someone else’s problem now outspends the chip giant that powers it.

Money talks, models listen. The next chapter of AI will be written partly in Washington, and the filings show exactly who is holding the pen.