California Governor Gavin Newsom has signed an agreement with Anthropic, granting state government agencies exclusive access to the AI assistant Claude. Under the deal, state agencies, as well as municipal and county authorities, will receive a 50% discount on the use of the tool.

In addition to the reduced price, the agreement includes free employee training, technical support, and the involvement of Anthropic developers in configuring workflows to meet the specific needs of government bodies. As Newsom emphasized, AI should not replace human labor in government but is intended to accelerate work, improve task efficiency, and enhance outcomes for California residents.

Claude becomes the first AI tool for performing work tasks available to all state agencies through a single portal of shared IT services, SITeS, operated by the California Department of Technology. According to the governor's office, Anthropic's product is already being used in several government institutions for customer service, optimizing internal processes, and cybersecurity tasks, including code scanning and fixing.

A Systematic Approach to AI Implementation

This agreement is part of Newsom's consistent strategy to integrate artificial intelligence into the public sector. As early as 2023, he tasked state agencies with exploring the potential applications of generative AI, and in 2025, he signed an executive order on measures to improve government efficiency. In March 2026, the governor also ordered an update of AI procurement rules to account for civil rights, privacy, and risks to employees.

Background: New Models and Restrictions

Recall that on June 9, Anthropic released two versions of models from the same family. Fable 5 was described as a Mythos-class solution but safe for general use, while Claude Mythos 5 is a "private" base model with relaxed restrictions in certain areas. However, just a few days later, developers restricted access to these tools for foreign nationals following a directive from the U.S. government. Some media outlets linked this to the use of the models by a group associated with China. At the end of the month, Anthropic accused operators linked to Alibaba of distilling Claude.

My comment: California is betting on AI as a tool to enhance operational efficiency rather than replace staff. However, against the backdrop of Anthropic's recent restrictions on foreign users and accusations of distillation, this partnership looks like an attempt to secure loyalty in the domestic market. The question is how transparent the criteria for using Claude in the public sector will be, especially in the context of cybersecurity and the protection of citizens' data.