Dealing with AI-assisted FOIA requests: ICO publishes new guidance

Public authorities are increasingly encountering information requests made under the Freedom of Information Act 2000 (“FOIA”) that appear to have been generated using AI tools. As AI becomes a standard tool for such correspondence, many public authorities are finding that AI-assisted requests are longer, broader and less focused than before, often making it harder to identify what information is actually being sought.

In response to this trend, the Information Commissioner’s Office (ICO) recently published new guidance on Freedom of Information (FOI) and Artificial Intelligence (6 May 2026). The guidance provides a practical framework for responding to the challenges commonly associated with AI‑generated FOIA requests, including overly broad or complex requests, factual inaccuracies and references to fictitious case law or ICO decisions.

Responding to AI-generated FOIA requests

The ICO is clear on one key point. Using AI does not, in itself, necessarily make a request invalid. FOIA is technology-neutral, and the existing FOIA framework is generally capable of dealing with AI-generated requests in the same way using any other technology (e.g. requests made on social media). Public authorities should continue to respond to AI-generated requests in the usual way, unless they fail to meet the requirements under section 8 FOIA. 

The guidance highlights that requests may still be valid even where they include inaccurate information or refer to non‑existent case law or ICO decisions. In such cases, the ICO encourages practitioners to politely inform a requester if AI has complicated the request or contains inaccuracies. The ICO frames this as part of the public authorities’ statutory duty under section 16 FOIA to provide advice and assistance to requesters. Rather than refusing requests outright, public authorities are encouraged to help requesters understand the issues and, where possible, to refine or correct the request.

Applying the existing FOIA tools 

Rather than reinventing the wheel, the guidance emphasises the continued relevance of existing FOIA principles and tools when dealing with AI‑assisted requests, including: 

• seeking clarification where AI makes the request unclear or more complex, or fails to clearly identify the information sought;
• applying the cost limit when a request or part of a request is so broad that it exceeds the cost limit, while working collaboratively with requester to narrow scope; and
• relying on section 14 refuse a request where AI is being used to generate repeated, burdensome or vexatious requests.

To support this approach, the guidance includes example wording that authorities may choose to include on their websites, encouraging requesters to use AI tools carefully and to ensure that requests are clear, accurate and proportionate.

Commentary

The guidance adopts a measured and pragmatic approach, integrating AI-related challenges into the established FOIA framework rather than creating a separate set of rules. For public authorities, the message is to continue to apply familiar FOIA principles - clarity, proportionality and good record‑keeping.

The guidance also underscores the importance of ensuring that practitioners are trained appropriately and understand what FOIA requires, and what it does not. In that sense, the guidance is less about AI itself, and more about reinforcing what good FOIA handling should already look like in practice.

For queries or advice on the content of this article, please contact Hamish Corner, Lucy Pegler, Amanda Leiu or a member of Burges Salmon's Commercial & Technology team.