The Department of Homeland Security (DHS) has issued new guidance prohibiting the use of commercial generative AI tools such as ChatGPT. 

In the memo to employees, DHS Chief Information Officer Antoine McCord instructed component technology offices to begin restricting access to generative AI tools and directed staff to use internal alternatives.

He also noted that earlier guidance — deemed “outdated” and too narrowly focused on commercial tools — has been removed from the department’s list of IT management directives.

This marks a shift from the department’s previous policy, which had conditionally permitted the use of commercial generative AI systems and signals a move toward leveraging internally developed technologies.

Earlier this year, the department launched an internal chatbot called DHSChat. According to Michael Boyce, Director of the DHS AI Corps, the chatbot enables employees to carry out routine tasks more efficiently, such as summarizing complex documents and reports, generating computer code and automating repetitive activities like data entry. 

Back then, he said that with this new tool, thousands of employees will be able to safely and securely harness generative AI capabilities using non-public data.

During the same time, DHS also released the “Playbook for Public Sector Artificial Intelligence Deployment”, a guide to help government officials enhance service delivery through the responsible and effective use of generative AI technologies.

The playbook offers actionable steps for state, local and federal officials and examples of how DHS applied and learned these principles in its own GenAI journey. 

DHS believes by implementing the recommendations,  other government agencies and organizations can build a robust foundation for AI deployment, enhance internal capabilities and ensure responsible and effective use of AI technologies.