Anthropic launches $15M cyber defense program for state, local, tribal and territorial governments

Anthropic announced the launch on Thursday of a $15 million cyber defense program aimed at helping state, local, tribal and territorial governments close funding gaps and protect vulnerable assets.

The artificial intelligence giant is committing up to $15 million in total Claude credits across the program, which are the cost per usage units Anthropic assigns to its AI services. The company said that smaller state, local, tribal and territorial governments can apply for up to $100,000 in credits, while additional program slots may be made available for large local government entities that include special districts supporting critical infrastructure like power, water, transit and port authorities.

These credits can be used to find and fix vulnerabilities using Anthropic’s publicly available models, like Claude Opus 4.8, Claude Security and Claude Code, as well as some of the company’s cyber defense runbooks and Claude “skills” — Anthropic’s term for workflow templates.

Participating governments can also receive hands-on support alongside other SLTT government tech leaders in a cohort, and weekly training and support from technical experts on Anthropic’s applied AI team. The company also noted that this program is distinct from Project Glasswing, the company’s restricted-access defensive cybersecurity initiative that allows select researchers to get their hands on the powerful Mythos model via Claude Mythos Preview.

The program comes at a time when states and local governments are struggling to close gaps left by cancelled federal funding and programs under the Trump administration, such as the State and Local Cybersecurity Grant Program, as well as the Multi-State Information Sharing and Analysis Center.

But state and local governments have already had significant cyber wins with Claude, the company said, claiming that one state’s security team recently found more than 40 vulnerabilities in its systems using Anthropic’s products in a matter of hours. California and Texas have already signed on to join the program.

“Having industry and government working together to test and validate what’s possible is preparing for the future—together,” Vitaliy Panych, California’s chief information security officer, said in a statement. “The reality is that bad actors will move quickly to operationalize AI against public institutions, and California defends more people and more critical systems than almost any government in the country. Frontier AI developers partnering with government is how we build the right tools to stay ahead of threats.” 

Texas Cyber Command Chief TJ White, VADM (ret), said that although there “remains a lot of uncertainty” surrounding fully mature AI capabilities and the effects on cybersecurity,  “we do know that any number of bad actors will try to operationalize this, bringing a degree of disruption which will be difficult to deal with at speed.”  

“This teamwork between Texas Cyber Command and Anthropic ensures we aren’t just reacting to potential disruptions, but building the agility needed to outpace emerging threats,” he added. “Partnerships like these are how we get ahead, and stay ahead.”

Michael Lai, Anthropic’s lead on AI for state and local government, said this program will hopefully help expand the high-stakes work that state and local governments do to protect their critical systems — especially as AI reshapes the sophistication of some attacks.

“State and local governments run the elections, benefits systems, and emergency services Americans rely on every day,” Lai said in a statement. “This program provides them with Claude credits and hands-on support so they can find and fix vulnerabilities. This is a natural extension of our work with the federal government to expand American cyber defense.”

The first cohort kicked off Thursday, with representation from a number of state and local governments. More cohorts will follow on a rolling basis through the summer, the company said.