Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Technology Policy topic
No spam. Unsubscribe anytime.
Select Board reviews draft AI policy; asks staff to add disclosure language and return revised draft
Summary
The Select Board reviewed a draft municipal AI policy and discussed guardrails, prohibited inputs (sensitive data, legal/medical advice), disclosure of AI use, and whether a broader data-privacy rule is preferable. Staff was asked to incorporate disclosure language and revised prohibited-data wording and return the policy on the next consent agenda.
Town staff presented a draft Artificial Intelligence policy on June 1, recommending rapid adoption alongside staff training and guardrails to protect sensitive town data and ensure staff accountability for AI‑assisted outputs.
Assistant town staff framed the policy as a balance: encourage staff use of AI to improve productivity while preventing insensitive or unauthorized disclosures. The draft included prohibited uses (entering prohibited data, seeking legal or medical advice, and personal use during work time) and required that "all output generated by AI must be reviewed by a town employee before release."
Board members debated whether the town should instead adopt a broader data‑privacy policy rather than an AI‑specific rule, and several members pushed for a clear disclosure requirement when generative AI materially assists in producing public documents. Dr. Heather Goldstone urged explicit bans on entering sensitive data into unconstrained models and asked staff to broaden the prohibited‑data language to include any online tool not approved by IT.
Members agreed to send the draft back to staff for revision: add a short disclosure mechanism for public documents (and a suggested standard email signature note), clarify prohibited data and allowed constrained tools, and return the revised policy on the next consent agenda with suggested training steps.
Action: Board requested staff to produce a revised AI policy incorporating disclosure wording and broader prohibited-data language, to be brought back on the next consent agenda.
