The promise of AI as a productivity multiplier has hit a hard wall in UK public administration, where a phenomenon dubbed "AI clog" is overwhelming civil servants. Local authorities, MPs' offices, and transport bodies are reporting that half of their incoming correspondence is now generated by chatbots, turning routine tasks like parking appeals and bin collection complaints into exhaustive legal challenges. This surge has transformed the role of public sector caseworkers, who previously handled two-page letters, into sifters of massive, often hallucinated, AI-generated documents.

The Scale of the Inundation

The volume of work has increased dramatically, with Deborah Evans, chief executive of Lawyers in Local Government, noting that while complaints used to average two pages, the majority are now ten times longer. The extreme cases are staggering, with Evans citing one complaint reaching 200 pages. This isn't just about length; the content is often bloated with irrelevant case law from different jurisdictions, such as Canada, due to AI hallucinations. Hollie Wickens, a senior caseworker in a Labour MP's office, confirms that half the emails her team receives are AI-generated, making requests significantly more taxing despite the automation promises made by tool developers.

Infrastructure Strain and Resource Misallocation

Public bodies are facing a critical resource mismatch because their funding does not scale with complaint volume. Evans points out that local authorities operate from a single pot of money, meaning every hour spent parsing AI-generated text is an hour not spent on actual service delivery. The impact is visible across sectors: Transport for London recorded a historic 4,769 Freedom of Information requests in 2025-26, a 5% increase over the previous peak, with a noted rise in appeals containing highly technical but inaccurate assertions. Similarly, employment tribunal claims opened between April and June soared by 28% year-on-year, contributing to a backlog of 70,000 open single claims.

Mitigation Strategies and the Machine-to-Machine Loop

In response, some organizations are attempting to stem the tide through structural changes. Wiltshire council is exploring AI tools to summarize long emails and draft initial responses, while MPs' offices like that of Graeme Downie have stopped responding to automated "click to email" campaigns entirely. The Information Commissioner's Office warns that even inaccurate AI-generated requests must be answered by humans, creating a bottleneck. However, the reliance on AI to filter AI creates a new risk: the loss of the human touch. Jamie Bartlett, a technology journalist observing the trend, argues that without stricter rules on what is accepted or the implementation of submission fees, public bodies will remain completely overwhelmed by the flood of synthetic text.

Key Takeaways

  • "AI clog" is now mainstream, with half of emails in some MP offices being AI-generated.
  • Average complaint length has jumped from 2 pages to over 20 pages, with extremes hitting 200 pages.
  • Employment tribunal claims rose 28% in Q2 2026, exacerbating a 70,000-case backlog.
  • Local authorities are not receiving additional funding to handle the increased administrative load.
  • Mitigation includes word limits, online templates, and in some cases, using AI to filter AI.

The Bottom Line

We built LLMs to help us write, but we forgot that writing is only half the battleβ€”reading is the bottleneck. Until we implement hard constraints on automated correspondence or charge for the privilege of spamming public servers, we are just automating the inefficiency.