Secrecy Concerns in Government Legal Agencies Adopting AI

Government legal agencies are adopting artificial intelligence (AI) to address staffing shortages and improve efficiency, writes Thomson Reuters. The 2026 Government Legal Department Report highlights growing optimism about AI along with the need for stronger governance, practical guidance, and safeguards to protect sensitive information. Nevertheless, confidentiality, data security, and accuracy concerns remain significant barriers.

Only one-third of government legal agencies use or plan to adopt legal-specific generative AI tools, and just 7% use any form of agentic AI. Nearly 80% of surveyed respondents worry about confidential data exposure, 73% cite AI accuracy and errors, and one-third express concerns about data retention and storage locations.

Shadow AI creates additional vulnerabilities. When employees use unauthorized public AI tools, they may inadvertently disclose protected information or expose it to security breaches. A survey including government professionals found that 27% had used external AI tools for work. Effective, approved alternatives, and clear usage policies are essential to reducing these risks.

Government agencies generally pursue AI through secure software-as-a-service (SaaS) platforms, customized third-party solutions, or internally developed and maintained systems. Costs rise across these approaches, often limiting agencies with smaller budgets to options involving greater external dependencies. Consequently, security planning must reflect the distinct risks of each implementation model.

These budget limitations can push agencies toward third-party AI solutions that require broader access permissions for employees and vendors, increasing exposure to credential-based breaches. As electronic breaches affecting court systems become more frequent and extensive, agencies must carefully evaluate technologies that increase access to sensitive records.

As government legal agencies are adopting AI, successful adoption requires balancing efficiency gains with confidentiality obligations. Agencies should strengthen cybersecurity, implement appropriate access controls, and establish clear policies. Collaboration among legal teams, support staff, IT teams, and leadership can support responsible adoption while protecting sensitive government information.