Government employees turn to public AI tools that don't meet government standards for security, privacy, or data reliability.
| Year added | 2026 |
|---|---|
| Status as of Spring 2026 | Pilot |
| Primary Agency | Central IT / CIO Office |
| Approach and Solution | State-developed generative AI assistant for CA government employees. Multiple leading AI models in one platform; grounded in CA.gov data. |
| Tags | |
| Who it Serves | Government staff |
| Area of Impact | Operations & Productivity |
| AI Technology Type | Chatbot, Document Processing, Process Automation |
| Level of Human Oversight | Routine use (no human review required) |
Alaska's Office of Information Technology (OIT) added AI enhanced Teams transcription and Microsoft 365 Copilot to the state's secure enterprise license, giving employees a vetted tool to draft, summarize, and prototype AI-enabled workflows inside the state's existing data perimeter.
The California Department of Technology (CDT) publishes an annual report inventorying high-risk automated decision systems used by state agencies. The 2025 initial filing reported no in-scope systems, a finding that drew public criticism.
The Indiana Office of Technology ran an eight-week Microsoft Copilot pilot across multiple agencies in fall 2025, then used the following months to address risks and gaps before rolling out Copilot Chat to all state staff. Any employee can request access after completing required training, and agencies can purchase the full M365 Copilot license on top of that. Training extends to the governor's cabinet secretaries, who attend sessions to learn the tool's capabilities, limits, and appropriate use.
Massachusetts announced an enterprise AI assistant powered by ChatGPT — through a partnership with OpenAI selected via competitive procurement — making it the first state to deploy the tool enterprise-wide. The assistant will eventually be available to nearly 40,000 executive-branch employees through a phased rollout beginning with the Executive Office of Technology Services and Security (EOTSS) and expanding to other agencies in the coming months, operating in a secure environment where inputs are not used to train public AI models. Employee use is optional and supported by EOTSS Privacy Office terms and conditions, agency-level points of contact, training programs, and an AI Assistant Knowledge Center; the tool is intended for tasks like drafting first versions of documents, research, content translation, and summarizing spreadsheets, and is explicitly not used to make eligibility decisions or to communicate with residents on behalf of the state.