California's governor announced on June 29, 2026 that the state will partner with Anthropic, an AI company, to bring Claude—a conversational AI tool—into state government agencies. This is the first time a U.S. state has made this kind of formal partnership with a major AI company.

The move builds on earlier steps Newsom took in April 2025, when California began using AI to tackle specific problems: highway traffic, road safety, and helping people apply for government benefits. Those were called the first AI deployments by any state. This new partnership is bigger—Claude will be available across multiple agencies rather than solving just one problem.

How Anthropic Positioned Itself

Anthropic didn't wake up one day and get handed this deal. The company spent roughly two years laying groundwork. In August 2025, it offered Claude to the federal government for $1 a month, removing a big barrier: complicated government purchasing processes. Anthropic also created a version of Claude designed specifically for government use, with extra security and privacy controls.

The company also made a smart regulatory move. In September 2025, it publicly supported a California law—SB 53—that would impose safety rules on advanced AI companies. Normally companies fight regulations that affect them. Anthropic's willingness to support one was unusual, and it signaled trustworthiness to the state government.

Before that, in March 2025, Anthropic also joined conversations with the Governor's Working Group on AI Frontier Models, helping shape the policy rather than ignoring it. All of this points to a strategy: help write the rules, then become the kind of company the state wants to hire.

What This Partnership Actually Means

Getting AI into state government is different from how most companies buy software. Government IT needs to know where data lives, keep detailed records of what AI does, control who can use it, and follow purchasing rules. Off-the-shelf AI doesn't usually handle these things out of the box. Because Anthropic already built a government version of Claude, some of that extra work goes away.

The announcement didn't specify which state agencies get Claude or what they'll use it for. That lack of detail is important. State agencies handle different types of sensitive information and use different computer systems. One agency might be ready to adopt Claude quickly; another might take years. A statewide partnership doesn't guarantee everyone will use it the same way or at the same speed.

Getting AI tools to actually work inside government offices takes longer than people expect. California's Department of Motor Vehicles has been upgrading its computer systems for years, which shows how slow government tech changes can be. That's true even when leadership wants to move fast.

But the bigger picture is becoming clear. Major AI companies—Anthropic, OpenAI, and Google—are deliberately moving into government. California, being the largest state economy and home to most of America's AI industry, is a test case everyone watches. If Claude works well in California offices, other states will likely ask for the same deal. If it runs into problems with security or rules, that tells people something too.

For Anthropic, this is the result of carefully building relationships and trust over two years. The real question is whether things go smoothly when the work actually starts.