UK public bodies struggle with longer AI-written complaints, staff say

A complaint that takes moments to draft can leave public staff with pages of work. In reporting published November 8, the Guardian describes UK councils and MPs’ offices struggling with lengthy correspondence they believe was written using AI. Staff say the tools are making complaints harder to handle, not just easier to send.

From a trickle to longer case files

Hollie Wickens, a senior caseworker in a Labour MP’s office, traces the change to 2024. What began as a trickle has become an “onslaught,” she told the Guardian. She estimates that half her office’s incoming emails involve AI in some form, making requests more complex and taxing to process despite her decade in the job.

Deborah Evans, chief executive of Lawyers in Local Government, describes a sharp increase in length. Complaints to local authorities once averaged about two pages, she says; the majority are now ten times longer. The largest example she knows of runs to about 200 pages. Her account describes staff experience, rather than a national measurement.

The extra work is not simply reading more words. Evans says chatbots can turn a complaint into something resembling a legal claim, citing statutes and cases that do not fit the relevant jurisdiction. A council may receive Canadian case law, for example. Invented material can add another problem for the people handling the complaint.

The requests still need answers

The Information Commissioner’s Office found that many public authorities received more requests in 2025. It warned that AI-generated requests can be repetitive, mix in unrelated comments or misdirected demands, and obscure the information actually sought. Yet requests that appear AI-drafted and contain inaccuracies still need a human response, the watchdog cautioned.

Transport for London received a record 4,769 freedom of information requests in 2025–26, 5% above its previous peak in 2023–24. That total counts requests overall; it does not measure how many were written using AI.

Its appeals provide a more specific warning. Committee papers cited by the Guardian recorded rising appeals and a “noted increase” in those that appeared AI-generated. These often contained highly technical but inaccurate assertions and assumptions. The distinction matters: TfL identified suspected AI use in appeals, not a verified AI share of its entire request volume.

In Westminster, the pressure is also landing on staff hours. Vonnie Sandlan, who heads Labour MP Graeme Downie’s office, said staff were exhausted and working beyond their paid hours. Lisa Gillmore, president of the GMB union’s MPs’ and peers’ staff branch, said casework had doubled or more in offices keeping their own figures.

Local authorities have one pot of money and don’t get extra money because of more complaints

Deborah Evans, chief executive of Lawyers in Local Government, speaking to the Guardian

Limits, templates—or AI replies

Responses described in the Guardian’s reporting range from restricting incoming material to using AI on the receiving end. They are at different stages: some limits are already operating, while Wiltshire Council is exploring automated help.

• Shorter submissions: Evans says some ombudsman services are imposing word limits or designing online templates that restrict attachments. These measures constrain the amount of material submitted rather than relying on identifying whether a chatbot wrote it.
• Fewer campaign replies: Sandlan’s office has recently stopped responding to automated “click to email your MP” campaigns. She says the decision is intended to let staff concentrate on constituents’ requests as they manage the heavier workload.
• AI-assisted handling: Wiltshire Council says it is exploring tools to summarise long emails and draft initial responses. That is a proposed use of AI to help staff process correspondence, not an announcement that complaint decisions will be automated.

The access question complicates those responses. Sandlan acknowledges that AI can help people who would otherwise struggle to draft correspondence. Evans, meanwhile, warns that using AI to sift through AI-written messages brings costs and risks “losing the human touch.” The same tools making it easier to ask for help are testing how institutions can keep providing it.