Anthropic and OpenAI Split Over AI Regulation
2 min readTwo of the biggest names in artificial intelligence now openly disagree about how their own industry should be governed. In July 2026, both Anthropic and OpenAI threw their weight behind state-level AI bills, yet the two labs split sharply over what national AI regulation should look like next.
A Federal Vacuum the States Rushed to Fill
For more than a year, the Trump administration has urged Congress to pass a law that would preempt state AI rules with a single federal standard. That effort has stalled. In the meantime, individual states pressed ahead. California passed S.B. 1053, New York advanced its RAISE Act, and Illinois enacted S.B. 315, which requires annual audits of high-risk AI systems.
The result is a fast-growing patchwork of state laws, and AI companies have had to decide whether to fight it or help shape it.
Where the Two Labs Agree, and Where They Don’t
Anthropic became the first AI company to endorse California’s S.B. 1053, arguing that the bill simply formalizes safety testing it already performs. It also backed New York’s RAISE Act and Illinois’s S.B. 315, and OpenAI endorsed the same measures, according to The Hill.
The agreement ends there. Anthropic wants each new state bill to add stronger safety duties, and it opposes any federal law that would override state rules unless that federal law is at least as strict as what Anthropic itself proposes. OpenAI is pushing the opposite way. Its chief of global affairs favors a “de facto” national framework built by aligning a handful of major state bills around a common baseline, an approach the company calls “reverse federalism.”
Why the AI Regulation Split Matters
The disagreement is about more than legislative tactics. A patchwork of tough state laws could force developers to meet the strictest rule in the country everywhere they operate, raising the safety floor along with the compliance burden. A single national baseline would be simpler for companies, but critics worry it could lock in weaker protections and freeze the rules in place as the technology races ahead.
With Congress deadlocked, the states have become the real battleground for AI policy. How this split resolves will help decide who sets the rules for the most powerful software ever built, and how quickly those rules can keep pace.