Carmarthenshire County Council has adopted a new policy governing the use of generative AI tools such as ChatGPT and Copilot.
The local authority said the policy has been created to ensure that any use of Generative AI within the council is "ethical, secure, and compliant with legal and regulatory standards".
It also said the policy was designed to prevent "automated biases", which some Generative AI tools have been accused of exhibiting.
Generative AI refers to AI products which can create new content, such as text, images, audio, or code, based on patterns it has learned from large datasets. This includes many easily accessible programmes such as ChatGPT, Copilot, Google's Gemini, and Bard.
The document states that users may use Generative AI tools for work-related tasks such as creating reports, emails, presentations, images, and customer service communications, as long as it is in line with the usage principles of the policy.
However, any information generated by Generative AI must also be reviewed and edited for accuracy prior to use, under the policy.
"If a user has any doubt about the accuracy of information generated by GenAI, they should not use it," it adds.
The document meanwhile warns users that Generative AI can "never be assumed to be accurate" as it can produce "hallucinations", which are outputs that appear plausible but are factually incorrect, misleading, or entirely fabricated.
It also notes that GenAI “can potentially produce inaccurate results, and outputs should be thoroughly verified for accuracy. Incorrect GenAI outputs can have substantial political, legal, and financial consequences for the authority. “
Examples cited include: “In 2021, a UK local authority faced legal challenges after relying on AI generated data that incorrectly identified properties for council tax reassessment, leading to wrongful tax increases.
“Another public sector organisation used AI to screen job applicants, which inadvertently introduced bias and led to discriminatory hiring practices.”
On data protection, the document says that users must not enter confidential and personal information into any public GenAI tools, as they are outside of the control of the council, "and information may therefore enter the public domain".
It listed examples of information that should not be entered into public GenAI tools, which included personal identifiers (e.g., names, addresses, NI numbers), sensitive case details, unpublished reports or draft policies, social care records, safeguarding reports, internal audit findings, and legal case notes.
The guidance also said the tools must be used ethically and not to produce discriminatory, offensive, or inappropriate content.
Any significant decisions taken that are assisted by Generative AI should meanwhile be "explainable and documented", according to the documents.
The policy was approved by the local authority's cabinet during a meeting on Monday (12 January).
All individuals, including employees, casual workers, agency staff, elected members, volunteers, external consultants or contractors, who are granted access to digital facilities by the council, must agree to adhere to the policy, the document states.
