Two AI labs that agree on absolutely nothing just lined up behind the same Senate bill.

Anthropic and OpenAI are in court against each other through proxies. They are racing to IPO. They are competing for the same Pentagon contracts. They are bidding on the same researchers and burning through billions in capital trying to put each other out of business. Their CEOs do not return each other's calls.

This week they did something almost unheard of in the AI industry. They publicly endorsed the same piece of legislation. The Workforce Transparency Act, introduced by Senator Mark Warner of Virginia and Senator Ted Budd of North Carolina, would require the federal government to actually start measuring what AI is doing to American jobs. Real data, on a regular cadence, public.

Anthropic backed it. OpenAI backed it. Politico confirmed both endorsements on April 30. Cue the eyebrow raise.

The political math is almost too clean.

Warner is a Democrat from Virginia. He chairs the Senate Intelligence Committee, knows the tech industry better than almost any sitting senator, and has spent two years pushing for AI guardrails that the industry would actually accept. Budd is a Republican from North Carolina, a Commerce Committee member, and one of the few GOP senators who has taken a substantive interest in AI policy without veering into culture-war territory.

A Warner-Budd bill is the kind of thing that can actually get fifty-something votes. Bipartisan. Limited in scope. Hard to demagogue against. The bill does not regulate AI. It does not slow down AI development. It does not impose a tax. All it does is force the Bureau of Labor Statistics, in coordination with other federal agencies, to start collecting data on how AI is changing employment, wages, hiring patterns, and job displacement. Numbers, on a schedule, made public.

If you read the bill cold, you might wonder why any AI company would even bother to weigh in. It is so narrow it barely registers. That is exactly why both Anthropic and OpenAI moved on it.

Both companies know the jobs data is coming. Better to shape it than fight it.

Here is the strategic logic. Every poll for the last six months has shown Americans of all political persuasions are worried about AI taking jobs. The Politico-Public First poll that dropped this week showed even Trump's own MAGA voters now favor AI regulation, with the jobs question polling as their top concern. The political pressure for somebody, somewhere, to start measuring AI's labor impact is going to keep building until it gets satisfied.

If the AI industry stays silent, the data collection bill that eventually passes will be written by senators who hate them. Probably Bernie Sanders and Josh Hawley, in some unholy left-right alliance, with the worst possible reporting requirements bolted on. Mandatory disclosure of every internal headcount projection. Real-time reporting of every layoff. Public dashboards that political opponents can use as a club.

Or the AI industry can pre-empt that by endorsing the cleanest, narrowest version of the bill they can find. Warner-Budd is that version. Bureau of Labor Statistics, standardized reporting, government-led, no individual company exposure. Anthropic and OpenAI just chose the lesser of two inevitabilities. That is what a smart lobbying shop does.

The pairing problem nobody at OpenAI wants to talk about.

There is a slightly awkward subtext to OpenAI's endorsement. Sam Altman has been Capitol Hill's favorite AI CEO for two years. He has been the calm voice in the hearing room. The one who asked for regulation. The one who flew to fifteen countries and met with regulators. He is also the CEO of the company most aggressively shipping AI products that automate white-collar work. ChatGPT has changed how knowledge workers operate. Codex changed how developers work. Sora is about to change how marketing teams work.

Endorsing a bill that measures all of that is, at minimum, a tacit acknowledgment that the measurements will not flatter OpenAI. The numbers, when they come out, will show real displacement in real industries. Customer service, paralegal work, copywriting, basic coding, image production, voice acting. OpenAI's products are already moving those needles. The data will say so in plain English.

OpenAI is essentially endorsing the production of evidence that may eventually be used against it in regulatory proceedings, antitrust investigations, and political campaigns. That is either masterful long-term positioning or a calculated bet that the alternative is worse. Maybe both.

Anthropic's angle is simpler.

Anthropic is the AI lab that has consistently positioned itself as the responsible adult. Constitutional AI. Responsible scaling policy. The CEO who actually answers tough questions in interviews. Backing a workforce transparency bill is on-brand to the point of being predictable. They get the press cycle, they get the policymaker meetings, they get to keep their reputation as the one in the room who is actually thinking about this.

It also costs them very little. Anthropic's customer base is heavily skewed toward enterprises building internal tools and developer-facing products. They sell less directly into the consumer-job-displacement bucket than OpenAI does. The data, when it lands, will probably be slightly kinder to Claude than to ChatGPT. They can afford to support measurement that they expect to look reasonable.

The political opening here is real.

If the bill passes, and the Warner-Budd framing makes that a real possibility, this becomes the first piece of meaningful federal AI legislation since the Biden executive order. The Trump White House has been allergic to AI regulation, but a bipartisan bill that does nothing more than count things is hard to veto. Especially when the Republican co-sponsor is from the South and the data shows MAGA voters want it.

Watch the markup process. If Warner-Budd starts to move out of committee, expect a stack of amendments from senators who want to use it as a vehicle for harder regulation. That is when the Anthropic-OpenAI alliance gets interesting. They will be on the same side, lobbying to keep the bill narrow, fighting off amendments that turn it into something they did not sign up for. Two AI labs that hate each other, working in tandem on Capitol Hill. The trade press is going to enjoy that.

What this signals for the rest of 2026.

The AI lobbying playbook is shifting. For two years it was about preventing regulation. The next two years are about choosing the regulation that is least bad. Both labs just publicly committed to that pivot. They are no longer pretending the rules will not come. They are trying to write the version of the rules they can live with.

If you work in AI policy, this is the moment the conversation matures. If you work at an AI lab, the comms team's job just changed. If you work in Congress, you have a model now for how to get an AI bill across the line. Pick a Republican from the South, pick a Democrat from a swing state, keep the scope tight, and let both AI labs save face by endorsing it.

The era of unanimous AI industry resistance to federal oversight just ended. It ended with a press release and two awkward joint statements that almost nobody noticed. Watch for more of these. The labs have read the polls. The dam is leaking. They are choosing where the water flows.