The widespread adoption of generative AI is reshaping the way individuals interact with public bodies. From Freedom of Information (FOI) requests to formal complaints directed at local authorities and regulators, AI is altering both the volume and character of submissions that public bodies must handle while many organisations are also exploring whether AI can assist in managing that increasing workload. This article sets out the key opportunities, legal risks and practical considerations for public bodies.
A genuine opportunity
It is important to acknowledge the positive dimension of this shift. Composing a formal complaint or an FOI request has historically been daunting for many individuals, and procedural complexity has deterred people with legitimate grievances from pursuing them. Generative AI materially lowers that barrier, assisting individuals to articulate concerns clearly, identify relevant legal provisions and navigate processes that might otherwise feel inaccessible.
Well-structured complaints benefit public bodies too, supporting more efficient dispute resolution and helping identify systemic issues earlier. However, those benefits come with pressures that require careful policy development and operational planning.
Emerging pressure on public bodies
The volume of FOI requests received by public bodies has grown considerably in recent years, and the accessibility of generative AI, which allows text to be produced instantly and at minimal cost, is certainly a contributing factor. Public bodies already operating under budgetary constraints face real difficulty in absorbing this increase whilst maintaining statutory response timescales.
Beyond volume, the character of AI-assisted submissions presents its own challenges. Such submissions tend to be longer, more formal and more likely to reference legal authorities or threaten litigation, even where the underlying issue is relatively straightforward. The risk is that genuine issues become obscured within dense and repetitive prose, making it harder to identify the core complaint and provide a meaningful response.
Accuracy, hallucinations and the risk of error
One of the most significant risks associated with AI-generated submissions is the reliability of the output. Generative AI is capable of producing "hallucinations" - plausible-sounding but factually incorrect information. In the context of complaints and requests, this can include references to non-existent case law, mischaracterised statutory duties, misattributed decisions or conflated facts. The recent decision in Ayinde v London Borough of Haringey, in which a party was found to have cited non-existent or materially misstated authorities generated using AI tools, serves as a timely illustration of the consequences of failing to verify AI outputs.
For public bodies, this creates a dilemma: they must engage carefully and fairly with submissions while avoiding reliance on flawed factual or legal premises.
AI Responding to AI: Opportunities and governance risks
Some public bodies are now deploying AI tools to categorise, triage and assist in drafting responses to complaints and information requests. Used appropriately, such tools can offer significant efficiencies, particularly in high-volume administrative environments. However, they also carry important governance and accountability risks.
Public bodies must reach decisions that are lawful, rational, procedurally fair and grounded in an accurate assessment of the facts. Without adequate human oversight, clear governance frameworks and robust audit trails, AI-assisted decision-making risks compromising those requirements - particularly where complaints involve safeguarding issues, equality duties, or fundamental statutory rights.
There is also a broader reputational consideration. Public confidence may be undermined if individuals believe complaints are being assessed primarily by automated systems rather than properly considered by accountable decision-makers. Transparency regarding the role of AI in decision-making processes is therefore likely to become increasingly important.
Key legal considerations
The increasing prevalence of AI in complaints handling does not alter the underlying legal framework. Public bodies remain bound by the following:
- Procedural fairness: Undisclosed or inconsistent application of AI tools in assessing complaints could give rise to arguments of procedural unfairness.
- Legitimate expectations: Published complaints procedures may create legitimate expectations as to how processes will be conducted; departures must be justified.
- Rationality: Decisions based on AI outputs containing errors remain the responsibility of the human decision-maker and are susceptible to rationality challenges.
- Proportionality: Targeted policies, such as word limits or AI-use declarations, may be lawful and appropriate, but must be applied consistently and transparently.
- Equality: Any new policy must be assessed for its potential impact on those who rely on AI tools to overcome communication barriers, including individuals with disabilities or those for whom English is not a first language.
- Data protection and confidentiality: Where AI tools process personal data or sensitive information, public bodies must also consider their obligations under data protection legislation, including transparency, security and lawful processing requirements.
Practical steps
Public bodies should consider the following:
- Audit existing complaints and request processes to identify where AI-generated content is already placing strain on resources;
- Develop or update AI policies that address both external submissions and internal AI use;
- Train staff to identify AI-generated content and understand the risks of inaccuracies and hallucinations;
- Establish clear governance and oversight frameworks before utilising any AI tools in complaints handling;
- Conduct equality impact assessments in respect of any new policies or tools introduced; and
- Maintain clear records of where AI tools have informed decision-making or drafting processes, in order to support transparency and accountability.
Conclusion
Generative AI is likely to remain a permanent feature of the complaints and information-request landscape. While it has the potential to improve accessibility and efficiency for both individuals and public bodies, it also introduces new operational, legal and governance challenges. Public bodies that respond proactively - through clear policies, meaningful human oversight and transparency in the use of AI - will be better placed to manage the increasing demand while continuing to meet their public law obligations fairly and effectively.
