Centralia puts guardrails on city staff using generative artificial intelligence

Staff at the City of Centralia have some new guardrails in place around the use of generative artificial intelligence.

The Centralia Board of Alderman approved the new rules at its Monday meeting, as well as a policy form city staff must sign outlining when AI use is appropriate and when it's not. The policy notes that AI tools like ChatGPT or Google Gemini can be useful for drafting, research and other routine work, but they can also produce inaccurate information and create security concerns.

City documents showed the city believes this policy will not only provide a process to address improper AI use, but also give them a baseline to adjust as AI advances.

The policy says city employees may only use generative AI models for city business only when approved by their supervisor. Authorization may be employee specific, position-based, department-wide, project-specific or as appropriate. Employees still remain responsible for reviewing and verifying the AI-generated work.

The city also put in guardrails for which city documents and information cannot be input into the AI models. The banned material includes law-enforcement-sensitive, protected and confidential documents along with information tied to personnel, utility customers and cybersecurity.

Any AI-related records used for city business will remain subject to Missouri's Sunshine Law and records-retention requirements.

Moving forward, the approved policy will be distributed to city employees for them to sign. Supervisors will communicate permitted uses and restrictions within their departments.The city said identified no financial impacts.