From Office of the Information Commissioner (OIC)
Published on
Last updated on
This policy is intended to help people who are making submissions to the Office of the Information Commissioner to understand how Artificial Intelligence (AI) tools can be used effectively.
Our role is to conduct reviews of decisions made by FOI bodies on FOI requests. As a result, the most important information that you can provide in relation to your case is a clear and simple description of what has happened in your own words.
AI tools can be useful in helping to organise your thoughts or to improve grammar, etc. However, our experience has shown that AI-generated correspondence can be inaccurate and can contain large volumes of material which is irrelevant in the context of the service we provide. Longer, AI- informed documents often lack the clear information we require in order to assess the issue involved. Instead, they often provide a substantial amount of additional information which does not help us deal with your case and, generally, these cases take longer to deal with.
Our advice to you is that the best information you can provide are the answers to the specific questions we set out in our webform, most importantly, in your own words. This helps us to best understand what exactly has occurred and what role we can play in assisting you.
AI tools can produce inaccurate, misleading information or unnecessarily voluminous information. While they can be helpful, their use also carries risks such as:
Inaccurate statements of facts: AI may re-draft the factual information in your correspondence in a way that is no longer accurate. AI may also suggest the inclusion of additional information which is not applicable in your case. This may result in your cases including inaccurate statements of what happened.
Inaccurate legal information: AI may generate content that appears relevant and authoritative, but is incorrect or irrelevant to Irish law and our procedures.
Erroneous references: AI might produce references to laws, policies or decisions that do not actually exist. This is called “AI hallucination”.
Sending us in the wrong direction: AI may produce large volumes of information which may not be relevant to how we handle your case. This could result in the information that you deem important being overshadowed by AI generated material.
Delay: Information submitted us is examined to determine its relevance. Examining large volumes of irrelevant information will delay the progression of your case.
Rejection of your submission: In cases where the substance of a case cannot be readily determined, or where significant time is taken up to examine large volumes of information submitted, you may be asked to resubmit your case, in your own words, in order for us to proceed.
If you choose to use an AI tool to help you correspond with us:
• You should confirm the accuracy of the contents.
• Check all factual statements carefully to ensure that it reflects what actually happened in your case.
• Check all references to policy or law against reliable sources or documentation.
• Do not include material you do not fully understand or cannot explain to us.
• Be aware of the privacy risks of inputting your sensitive data into AI tools that may retain or process your data for other purposes.
• It would be useful if you inform us if you are using AI. If you have, then you should consider making a statement to that effect, for example: “Parts of this correspondence were drafted using an AI tool. I have reviewed and confirmed the accuracy of the content.”
Final note: You are responsible for the accuracy, relevance and honesty of your documents, even if you used AI to help prepare it.