Pocatello CIO proposes brief employee AI policy; council urges coordination with state

Chris Sorensen, the city’s chief information officer, presented a short draft artificial intelligence policy at a Pocatello City Council work session and asked the council for direction on adding it to the employee handbook.

Sorensen told the council the draft is deliberately small — three bullets that authorize generative AI for employees, require validation of AI output before use in city decision-making, and direct IT to develop an operational guidance directive. “Generative artificial intelligence is authorized for use by city employees to aid in the accomplishment of their assigned duty and mission,” Sorensen read from the draft. He added that employees “must always validate the accuracy of the AI output” before publishing or relying on it.

The policy is intended as a baseline while the city develops department-level guidance. Sorensen said the city has received departmental requests to use AI tools ranging from generative chat engines to biometric and surveillance technologies. He also warned of risks such as deepfakes, phishing and “data hallucination,” citing recent national stories in which AI-produced citations or facts were fabricated. “The onus of the responsibility for using that goes back to the user,” he told the council.

Why this matters: council members said a brief, enforceable policy would give staff guardrails while technology evolves, but they pressed for more clarity about platform choices, cybersecurity and state law. Councilmember Cheatham pressed a procedural point: “I think if we’re gonna adopt public policy for the city, any changes in a policy needs to come before this council at every step.” Other council members asked staff to contact state officials and confirm whether the state intends to mandate a single AI platform for agencies.

Council reaction and points of debate

- Platform standardization: Several council members raised the possibility of standardizing on a single AI platform so IT staff can become experts and departments have consistent training. One council member reported receiving a notice that the state legislative services office is considering choosing a single AI platform for state use. Other members warned that over‑restricting platform choice could hinder the city’s ability to procure software that includes AI features built into new products.

- Scope and enforcement: Sorensen and members agreed the draft should not attempt to regulate all uses of AI at the outset. Instead, they described a two‑tier approach: a short handbook policy to assign responsibility to employees, and a living operational‑guidance directive developed by IT (to be updated as technology and standards evolve). The operational guidance would cover acceptable practices, training and security measures and would be updated without returning to council monthly.

- Verification and ethics: Council members discussed how employees should document use of AI and whether AI‑generated creative work should be labeled. Sorensen noted legal and ethical issues are evolving and that guidance will address when crediting or disclosure is appropriate.

What staff will do next

Sorensen said staff will revise the draft based on council feedback, reach out to state contacts for clarity about any statewide platform decisions, and develop the operational guidance directive. If the council directs staff to proceed, IT and human resources will bring a final handbook amendment to a formal council meeting through HR for adoption.

Ending

Council members signaled general support for a short, enforceable policy that places responsibility on employees to verify AI output and for a separate, flexible operational guidance document. Staff will return with a revised draft and more information about the state’s intent on platform choices before a formal adoption step.