Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Ai Policy topic
No spam. Unsubscribe anytime.
Port Orchard staff seek council sign‑off on citywide AI policy after yearlong committee work
Summary
City staff briefed council on a year of AI policy work, recommending a resolution adopting an AI use and procurement policy that requires human review of AI outputs, product vetting, prompt logging where feasible, and staff training; resolution is slated to return to council for consideration next week.
City staff presented a proposed municipal artificial‑intelligence policy at the Port Orchard city council work study on March 17, recommending the council consider a formal resolution next week to adopt rules for AI procurement, use, transparency and oversight.
Shawn Dum, the city's IT manager, told the council that an interdepartmental AI committee formed after state guidance and first met in April of last year to educate staff, review case uses and draft policy language. "Our goal there was really about education," he said, describing a cross‑departmental committee that produced successive drafts and a 28‑page acquisition guideline for vetting AI tools.
Staff described several core features of the draft policy: a vetting and procurement checklist, an approved‑products registry, training for users, and a human‑in‑the‑loop requirement so subject‑matter experts must QA AI‑produced outputs before material is finalized. Dum said Microsoft Copilot is currently the only approved product because it runs in the government cloud and can log prompts and outputs.
The council also discussed public‑records implications. Charlie, the city attorney, summarized recent state guidance: when AI interactions "relate to the transaction of public business" both prompts and outputs may qualify as public records, though some AI‑generated search interactions can be treated as transitory if staff promptly delete caches. "If you have good records hygiene you're deleting them on a regular basis," he said, while cautioning that outputs used to draft ordinances or policies would have different retention requirements.
Council members pressed staff on privacy and vendor chains; staff said their vetting form will limit uploading non‑public data to vendors and that any procurement of higher‑risk products would trigger full committee review and pre‑launch training. Staff recommended adding a footer or attestation on documents that used AI and were human‑reviewed.
Staff asked the council for direction to bring back a formal resolution for adoption. The council signaled support for proceeding, with staff noting this will be an initial policy subject to adjustment after early deployment and evaluation.
The council is scheduled to consider a resolution adopting the AI policy at its next regular meeting; staff said they will circulate final draft materials and training plans before that meeting.
