More than $2 million in campaign donations flowed from the largest AI companies to state attorneys general committees since the start of 2025, according to IRS disclosures reviewed by Sludge. The donors include OpenAI, Google, Meta, Amazon, Anthropic, and Elon Musk's xAI. The recipients: the very AGs who are currently investigating these same companies over AI-generated child safety incidents, deceptive safety claims, and deepfake content. For AI founders watching from the sidelines, the message is unmistakable. Political influence has become a line item in the AI compliance budget.
More than half of the $2 million arrived in just the past few months. Google, Meta, and Amazon each donated matching amounts to both the Democratic Attorneys General Association (DAGA) and the Republican Attorneys General Association (RAGA) in the second quarter of 2026: $250,000, $200,000, and $125,000 respectively. Anthropic made its first-ever donation to either group with a $250,000 contribution to RAGA on May 6. And xAI donated $25,000 to DAGA in May while simultaneously suing California Attorney General Rob Bonta and being investigated by his office over Grok-generated deepfake images of minors.
The Bipartisan Investigation Machine
The regulatory pressure on AI companies is not a partisan phenomenon. In August 2025, a bipartisan coalition of 44 state AGs led by Republicans from South Carolina and Tennessee alongside Democrats from Illinois and North Carolina sent a joint letter to OpenAI, Meta, Google, xAI, and other AI firms warning of grave concerns over children's interactions with AI chatbots. A follow-up in December, led by Pennsylvania Republican Dave Sunday and New Jersey Democrat Matthew Platkin, demanded stronger safeguards after reports of harmful interactions with minors persisted.
In June 2026, New York Attorney General Letitia James subpoenaed OpenAI for documents covering treatment of minors, advertising practices, and user safety protocols after a string of public incidents. The bipartisan nature of this pressure means AI companies cannot simply align with one party to reduce regulatory risk. Both sides are active, and both sides are receiving donations.
The xAI Paradox: Litigation and Donations
The most striking case study is xAI. In December 2025, the company sued Bonta in federal court to block enforcement of California's AB 2013, a law requiring AI developers to disclose their training data. xAI argued the disclosure requirements amounted to an unconstitutional taking of trade secrets and compelled speech. Weeks later, on January 16, Bonta sent xAI a cease-and-desist letter demanding the company stop distributing sexually explicit deepfakes generated by its Grok model. Bonta opened an investigation after reports that Grok was producing nonconsensual sexualized images of women and girls, including content depicting minors, which he stated met the legal definition of child sexual abuse material.
In March, a federal judge denied xAI's request to block AB 2013 while the suit continued. Then on May 4, xAI made its first political donation: $25,000 to DAGA, the same organization whose members include the attorney general actively investigating the company. The optics would be hard to miss even if you tried.
What This Means for AI Founders
For solo founders and early-stage AI companies, this story reveals three hard truths about building in a regulated space. First, state-level AG enforcement is where AI regulation is happening right now, not in Washington. The 44-AG letter, the James subpoena, the Bonta investigation, and the bipartisan follow-ups all operate at the state level. Federal AI legislation remains stalled. If you are an AI founder building anything involving minors, content generation, or data disclosure, you should expect state AG attention long before the FTC or DOJ shows up.
Second, the donation patterns reveal a playbook that only works at scale. Writing $250,000 checks to both DAGA and RAGA is not a strategy available to a startup with less than $1 million in ARR. But understanding that the biggest AI companies are investing in political relationships as a hedge against enforcement gives you context for your own risk profile. If you cannot afford the donation route, your compliance posture needs to be bulletproof from day one.
Third, the xAI situation is a warning about the risks of building generative AI products without robust safety guardrails. Grok's deepfake generation capability triggered a direct investigation, a cease-and-desist, and a lawsuit all in the span of six months. If you are building image generation, voice cloning, or any content creation tool that could be abused, the margin for error at the state AG level is essentially zero. A single compliance failure can trigger a multi-state investigation that drains your runway.
What Founders Need to Do
The regulatory environment for AI at the state level is only going to intensify. Here is a practical checklist for founders who want to stay ahead of AG scrutiny. First, document your safety testing process thoroughly and prepare it for AG-level requests. The bipartisan 44-AG letter specifically demanded information about how companies test their models for child safety risks before deployment. Second, designate a state-level compliance point person who tracks AG activities in states where you operate or have users. Third, review your content moderation and abuse reporting systems with the assumption that an AG investigation will test them under adversarial conditions. Fourth, understand that donating to AG associations is a defensive play for large companies, not a substitute for safety engineering. For early-stage founders, the best defense is a product that demonstrably does not cause harm.