Policy
Anthropic Doubles Midterm Spending To Push AI Regulation
Anthropic committed another $20 million to Public First Action, bringing its midterm spending to $40 million and widening Silicon Valley's political fight over AI regulation.
By Michael C ยท
Anthropic is putting another $20 million behind its preferred vision of AI regulation, turning a policy argument into an election-year spending campaign. The Wall Street Journal reported that the company has doubled its midterm spending to $40 million by committing the new money to Public First Action, a political group backing government-imposed safeguards for powerful AI models and transparency from developers about the risks those systems pose. The move widens a Silicon Valley split that is no longer confined to white papers, model cards, or conference panels.
The spending lands in a volatile policy environment. Anthropic chief executive Dario Amodei has argued for stronger oversight of frontier systems, while other technology figures and companies have warned that heavy regulation could slow U.S. innovation or entrench the largest players. The Journal reported that Public First Action contrasts with Leading the Future, a rival group backed by OpenAI and Andreessen Horowitz executives that favors a more industry-friendly approach aligned with the White House. That makes AI regulation a campaign issue, not only an agency issue.
The policy caution is basic. Corporate political spending can support a public-interest position and still advance a company's business interests. Anthropic says stronger safeguards are needed because frontier models can create serious risks. That view is plausible. It is also true that a stricter regime may advantage companies with large compliance teams, government relationships, and the resources to conduct expensive testing.
The regulatory fight is also shaped by recent history. Anthropic has faced disputes over model access and national security treatment, while OpenAI has had its own release-review tensions with Washington. Those episodes created an unusual political moment. The most powerful AI companies are asking government to create rules, while also fighting over what those rules should contain and who should enforce them. The public should not mistake industry agreement that regulation is coming for industry agreement on the public interest.
The election mechanism matters because AI policy is still structurally unsettled. Congress has not created a comprehensive federal AI regulator. States are moving on their own. Federal agencies are issuing guidance, procurement rules, and enforcement signals. Political groups can therefore shape candidate positions before the statutory framework hardens. Money spent now may influence whether future AI law emphasizes federal preemption, mandatory evaluations, disclosure, liability, open-model restrictions, or lighter-touch innovation policy.
Transparency will be the central test for both sides. Voters and lawmakers deserve to know who is funding a campaign, what rules are being advocated, and which companies would benefit. A group that calls for safety should identify the specific obligations it supports. A group that warns against overregulation should explain how it would handle cyber, bio, deception, privacy, and labor risks. Slogans about safety or innovation are not enough for a technology this consequential.
The spending also complicates criticism of government involvement in AI. Technology companies often object when policymakers intervene in product decisions. They are also spending heavily to shape the interventions that do happen. That is not hypocrisy by itself. It is how regulated markets form. But it does mean claims about neutral technical governance should be examined alongside campaign finance and lobbying strategy.
Smaller AI developers and open-source advocates have reason to be cautious. If the coming framework is written primarily by the largest labs and their political allies, it may produce burdens that smaller actors cannot meet. If the framework ignores the risks of increasingly capable systems, it may damage public trust and invite a later backlash. The hard policy work is to make obligations proportional to capability and deployment context rather than corporate size alone.
Anthropic's $40 million commitment is therefore not only a campaign finance story. It is a signal that AI companies now see political infrastructure as part of the competitive stack. They are building models, buying chips, hiring safety teams, negotiating with governments, and funding the politics that will decide the rules around all of it.
The public interest will depend on whether lawmakers can take money and arguments from the industry without outsourcing judgment to it. Anthropic may be right that stronger rules are needed. Its spending makes that argument louder. It does not make it self-proving.
Topics: Anthropic, AI regulation, campaign spending, Public First Action