Byron Shire Council is proposing restrictions on entering residents’ personal and health information into artificial intelligence tools, with exceptions requiring an approved privacy assessment or the individual’s consent.
The safeguards were added by councillors before a draft AI policy was released for public comment, alongside proposed requirements for human oversight of AI-assisted decisions and documents.
Councillors unanimously endorsed the amended policy for exhibition at their October 15 meeting. Submissions are currently listed as closing on November 17.
The policy would apply across Council, covering councillors, staff, contractors, consultants and third-party systems incorporating AI.
Under the amendment moved by Cr Elia Hauge and seconded by Cr Michelle Lowe, personal information would receive specific protection.
‘Personal and health information must not be entered into, or made accessible to, any AI tool,’ the resolution states.
Exceptions could be permitted following a privacy impact assessment approved for the specific use, with a register reported to Council’s Audit, Risk and Improvement Committee, or where the individual had consented.
Any approved use would have to remain consistent with the purpose for which the information was originally collected and comply with privacy legislation and Council’s data-handling rules.
The original draft also places responsibility for AI-assisted work with people.
It states that ‘a named person, not an algorithm, remains accountable for every AI-assisted decision or output Council relies on’.
AI-generated content and decisions would require human review before being relied upon, published or acted upon, with scrutiny scaled according to risk.
The draft requires assessments for bias before AI is deployed in processes affecting residents, and says its use must demonstrably improve services, efficiency or accessibility.
Disclosure would generally occur through website statements and legal notices rather than individual labels on every AI-assisted document. Basic editing for grammar, clarity or formatting would not require disclosure.
A separate safety and security plan would govern approved tools, data handling, procurement and responses to misuse or breaches.
Those technical controls could be updated without a Council resolution, with higher-risk changes requiring executive approval.
The documents do not identify which AI tools Council currently uses or establish that residents’ information has been misused.
The proposal comes against a regional backdrop of concern about AI and privacy, following last year’s disclosure that a Reconstruction Authority contractor uploaded Resilient Homes Program information to an unauthorised AI tool.
There is no established connection between that incident and Byron’s policy proposal, which remains subject to the current public consultation process.
For four decades The Echo has printed the stories some people loved, some people hated, and some pretended not to read. If you want us to keep telling the truth, the real truth, not the sugar-coated version. We’ll need your support to keep the presses rolling. 
