Blab AI
Maryland Expands AI Use Across 50+ State Agencies, Raises Oversight Questions
From human services to transportation, Maryland’s state government is now an AI‑powered machine. A state‑wide inventory mandated by Senate Bill 818 lists more than 50 agencies, boards, and offices that employ artificial‑intelligence tools. Among the top users are the Department of Human Services, Department of Transportation, Department of Health, and the governor’s office. The governor’s office reports using AI for public inquiries, data analysis, drafting documents, reviewing records, and streamlining internal operations.
Last month, the Department of Information Technology (DoIT) opened an AI Innovation Lab designed to give agencies hands‑on access to cutting‑edge tools, infrastructure, and expert guidance. The lab partners with Anthropic, whose Claude chatbot has already been deployed by Governor Wes Moore to answer budget questions and compare his administration’s spending cuts with those of previous governors.
Researchers describe Maryland as an aggressive adopter of AI in government. They point out that as AI becomes more embedded in public services, residents need to know when the technology is used, whether it influences decisions, and what recourse exists when mistakes occur. The inventory lists both agency‑developed tools and commercial products such as ChatGPT, Claude, Google Gemini, and Microsoft Copilot, but offers limited information on costs, benefits, or the impact on residents.
"AI can be a really helpful tool, but the big concern is that it takes the human out of the equation of judgment and decision‑making," said Doug Lombardi, a researcher at the University of Maryland Artificial Intelligence Interdisciplinary Institute. "Marylanders are going to be using AI whether they realize it or not." Lombardi added that agencies should clearly explain how, when, and why they use AI.
Henry Farrell, an AI, data and democracy scholar at Johns Hopkins University, warned that while AI can help residents navigate complex government programs, it can also reduce transparency. "By relying on automated systems to sort information, generate outputs or assist with analysis, agencies may make it harder for the public to trace how data was collected, how it was processed and how it ultimately informed a decision," Farrell noted.
The governor’s office has referred questions about AI oversight to DoIT, which manages Maryland’s AI governance framework. Nathan Miller, a DoIT spokesperson, said any AI‑generated content must be reviewed and approved by a state employee before use. "The State of Maryland is using AI as a support tool," Miller wrote in an email. "Responsibility for any official work product, public statement or budget decision rests with the responsible state officials and employees."
Maryland’s Responsible AI Policy, which applies to executive‑branch employees, requires additional review for AI systems considered high‑risk. The policy is intended to prevent harmful or inappropriate uses of AI and guarantees that residents retain the same avenues to appeal or challenge government decisions, regardless of whether AI assisted the underlying work.
In a recent move to strengthen oversight, Governor Moore announced the appointment of Michael Boyce, a former Department of Homeland Security and White House technology official, as a senior adviser for responsible AI. Boyce will coordinate Maryland’s AI strategy as agencies continue to expand their use of the technology.
The issue has already surfaced in Maryland courts. In 2025 the Maryland Appellate Court overturned a robbery conviction after finding prosecutors failed to disclose their use of facial‑recognition technology to identify a suspect. The court ruled that the delayed disclosure prevented the defendant from adequately challenging the reliability of the AI‑assisted evidence.
Prince George’s County Sen. Ron Watson, who serves on the Senate Education, Energy and the Environment Committee, said Maryland lawmakers have regularly discussed how agencies are using AI and what guardrails are needed as the technology becomes more common in government. Watson emphasized that lawmakers focus on preventing bias, protecting personal information and ensuring fairness when AI is used in government processes. He cautioned that implementation is the biggest challenge and that oversight cannot rely solely on legislation.
As Maryland expands AI use across government, researchers say the central test will not be whether the technology improves efficiency, but whether residents can still understand, challenge and appeal decisions that affect their lives. The state’s AI inventory, the Responsible AI Policy, the AI Innovation Lab and the appointment of a senior adviser represent steps toward that goal, but questions about transparency, oversight and the potential for high‑risk AI remain.
The AI inventory is publicly available at ai.maryland.gov and is updated regularly. The Responsible AI Policy and its implementation guidance are posted on the DoIT website. The AI Innovation Lab will continue to support agencies in prototyping software, identifying cybersecurity vulnerabilities and testing new AI applications.
Maryland’s experience illustrates the broader challenges facing states that adopt AI in public services: the need for clear disclosure, robust oversight, and mechanisms that allow residents to hold decisions accountable. The state’s ongoing efforts to address these issues will likely influence other jurisdictions as they navigate the growing role of AI in government.
Last month, the Department of Information Technology (DoIT) opened an AI Innovation Lab designed to give agencies hands‑on access to cutting‑edge tools, infrastructure, and expert guidance. The lab partners with Anthropic, whose Claude chatbot has already been deployed by Governor Wes Moore to answer budget questions and compare his administration’s spending cuts with those of previous governors.
Researchers describe Maryland as an aggressive adopter of AI in government. They point out that as AI becomes more embedded in public services, residents need to know when the technology is used, whether it influences decisions, and what recourse exists when mistakes occur. The inventory lists both agency‑developed tools and commercial products such as ChatGPT, Claude, Google Gemini, and Microsoft Copilot, but offers limited information on costs, benefits, or the impact on residents.
"AI can be a really helpful tool, but the big concern is that it takes the human out of the equation of judgment and decision‑making," said Doug Lombardi, a researcher at the University of Maryland Artificial Intelligence Interdisciplinary Institute. "Marylanders are going to be using AI whether they realize it or not." Lombardi added that agencies should clearly explain how, when, and why they use AI.
Henry Farrell, an AI, data and democracy scholar at Johns Hopkins University, warned that while AI can help residents navigate complex government programs, it can also reduce transparency. "By relying on automated systems to sort information, generate outputs or assist with analysis, agencies may make it harder for the public to trace how data was collected, how it was processed and how it ultimately informed a decision," Farrell noted.
The governor’s office has referred questions about AI oversight to DoIT, which manages Maryland’s AI governance framework. Nathan Miller, a DoIT spokesperson, said any AI‑generated content must be reviewed and approved by a state employee before use. "The State of Maryland is using AI as a support tool," Miller wrote in an email. "Responsibility for any official work product, public statement or budget decision rests with the responsible state officials and employees."
Maryland’s Responsible AI Policy, which applies to executive‑branch employees, requires additional review for AI systems considered high‑risk. The policy is intended to prevent harmful or inappropriate uses of AI and guarantees that residents retain the same avenues to appeal or challenge government decisions, regardless of whether AI assisted the underlying work.
In a recent move to strengthen oversight, Governor Moore announced the appointment of Michael Boyce, a former Department of Homeland Security and White House technology official, as a senior adviser for responsible AI. Boyce will coordinate Maryland’s AI strategy as agencies continue to expand their use of the technology.
The issue has already surfaced in Maryland courts. In 2025 the Maryland Appellate Court overturned a robbery conviction after finding prosecutors failed to disclose their use of facial‑recognition technology to identify a suspect. The court ruled that the delayed disclosure prevented the defendant from adequately challenging the reliability of the AI‑assisted evidence.
Prince George’s County Sen. Ron Watson, who serves on the Senate Education, Energy and the Environment Committee, said Maryland lawmakers have regularly discussed how agencies are using AI and what guardrails are needed as the technology becomes more common in government. Watson emphasized that lawmakers focus on preventing bias, protecting personal information and ensuring fairness when AI is used in government processes. He cautioned that implementation is the biggest challenge and that oversight cannot rely solely on legislation.
As Maryland expands AI use across government, researchers say the central test will not be whether the technology improves efficiency, but whether residents can still understand, challenge and appeal decisions that affect their lives. The state’s AI inventory, the Responsible AI Policy, the AI Innovation Lab and the appointment of a senior adviser represent steps toward that goal, but questions about transparency, oversight and the potential for high‑risk AI remain.
The AI inventory is publicly available at ai.maryland.gov and is updated regularly. The Responsible AI Policy and its implementation guidance are posted on the DoIT website. The AI Innovation Lab will continue to support agencies in prototyping software, identifying cybersecurity vulnerabilities and testing new AI applications.
Maryland’s experience illustrates the broader challenges facing states that adopt AI in public services: the need for clear disclosure, robust oversight, and mechanisms that allow residents to hold decisions accountable. The state’s ongoing efforts to address these issues will likely influence other jurisdictions as they navigate the growing role of AI in government.
