UK public bodies are struggling with a flood of complaints and requests written with the help of chatbots, in a problem staff have started to call AI clog. The Guardian reported on 8 November that MPs, local councils, schools, courts, lawyers and businesses are all receiving the same kind of correspondence: longer, more formal and far harder to deal with than anything they have seen before. Hollie Wickens, a senior caseworker in a Labour MP’s office with a decade of experience handling requests from the public, said about half of the emails her team now receives show signs of having been drafted by artificial intelligence.
Wickens said the change came fast. “It started as a trickle in 2024 and now it’s an onslaught,” she told the Guardian. The tools behind the AI clog are familiar ones: ChatGPT, which turns four this month, alongside Claude, Copilot and Gemini. Together they have handed ordinary people a free legal adviser and personal assistant, and a letter that once took an evening to research and write can now be produced in seconds.
The strain comes from the mismatch that follows. Writing got far easier, but reading and replying still take the same amount of work, and the organisations on the receiving end say the systems built to handle complaints were never designed for this volume. Researchers at the Hertie School in Berlin, the Cooperative AI Foundation and GovAI have given the pattern a more technical name, agentic flooding, and they warn that the sheer volume of machine-assisted messages risks overwhelming the institutions citizens rely on, and the AI clog is already doing that work in council offices.
Local councils feel the pressure most directly. Alessandro Georgiou, the council leader in Enfield, north London, told the Guardian that more and more residents are using AI to challenge council decisions, with a clear rise in appeals against parking fines. Deborah Evans, chief executive of Lawyers in Local Government, which represents 5,600 lawyers who work for local councils, said every part of the sector is reporting the same problems.
AI clog and the longer letters
Evans said the AI clog shows first in the length of what arrives. Before chatbots went mainstream, a complaint sent to a council ran to about two pages on average. Today, she said, most are ten times longer, and the most extreme case she has heard of runs to about 200 pages. The letters tend to be repetitive, padded with off-topic comments and demands aimed at the wrong place, and each one still has to be read in full by a person.
The Information Commissioner’s Office, the UK body that oversees information rights, has spotted the same AI clog. It said many public authorities received more requests in 2025, and the ones that look AI-generated tend to be long, repetitive, full of off-topic comments and demands aimed at the wrong place. The watchdog stressed that requests that look like they were drafted by AI and contain inaccuracies still need a response from humans, so staff cannot simply set them aside.
Transport for London logged a record 4,769 freedom of information requests in 2025-26, a rise of 5 percent on its previous peak. Councils mostly spot the AI clog because of message length, certain stock phrases and heavy use of bullet points, but suspicion is not proof, and a long letter can still hold a genuine complaint that deserves an answer.
AI clog and the tribunal backlog
Employment tribunals show how the AI clog adds to systems that were already under strain. Single-claim cases opened between April and June rose 28 percent on the year before, to 14,000, and about 70,000 single-claim cases remained open at the end of June. Tribunals had a backlog before chatbots arrived, and longer, more complex claims take more time to work through at each stage.
The cost of the AI clog lands on budgets that are already tight. Councils, MPs’ offices and courts pay staff to read, check and answer correspondence that a machine produced in seconds, and there is no simple filter that separates a genuine grievance from a padded one. A resident who uses a chatbot to write a parking appeal still has a case to be heard, and a council that misses a deadline still faces the consequences.
Nobody in the Guardian report offered a quick fix. The researchers who named the pattern agentic flooding argue that public bodies need new ways to handle machine-assisted correspondence, from better triage to clearer rules on what counts as a valid request. For now, the people answering the letters say the work keeps growing while the staff to do it does not, and Wickens put it plainly: what began as a trickle in 2024 is now an onslaught.