'AI Clog' Strains UK Public Bodies and Courts as Automated Complaints Surge
Generative AI tools are flooding UK councils, MPs' offices, and tribunals with lengthy, automated correspondence, sparking demands for technical filtering and structural reform.
Public institutions and private enterprises across the United Kingdom are confronting an unprecedented surge in artificially generated correspondence, a phenomenon researchers from the Hertie School, the Cooperative AI Foundation, and GovAI call "agentic flooding" and commentators term "AI clog." Generative tools such as ChatGPT—which turns four later this month—Claude, Copilot, and Gemini have lowered the barrier for citizens to write complex emails, formal complaints, and legal appeals. However, while these applications turn laypeople into self-taught administrators, public bodies and businesses report that the resulting volume of verbose and often inaccurate material is severely straining their operational capacity.
Escalating correspondence across UK public bodies
Local councils, public sector legal departments, and regulatory authorities are experiencing unprecedented backlogs due to lengthy AI-drafted documents. Deborah Evans, chief executive of Lawyers in Local Government (LLG)—representing 5,600 council lawyers—notes that traditional complaints to local authorities historically averaged around two pages. Today, the majority of AI-assisted complaints are ten times longer, with extreme instances reaching up to 200 pages.
Evans points out that while AI tools can make complaints sound formal and professional, they frequently introduce irrelevant statutes, references to incorrect jurisdictions such as Canadian case law, or entirely hallucinated facts. Nevertheless, guidance from the Information Commissioner’s Office (ICO) mandates that public authorities must still process and answer AI-assisted requests using human staff, even when those submissions contain inaccuracies or repetitive demands.
The strain is evident in specialized departments and local governance. Alessandro Georgiou, leader of Enfield council, reported a notable rise in residents utilizing AI to appeal council decisions, particularly regarding parking tickets. Meanwhile, Transport for London (TfL) handled a record 4,769 freedom of information (FoI) requests during the 2025–26 period, representing a 5% increase over its previous peak in 2023–24. TfL audit documents noted a clear rise in AI-generated appeals containing technical and inaccurate assertions.
Legal disputes and strained workplace operations
The impact extends heavily into employment relations and parliamentary offices. Hollie Wickens, a senior caseworker in a Labour MP's office with ten years of experience, observed that approximately half of incoming constituent emails now involve AI generation, making requests far more complex and time-consuming to resolve. Vonnie Sandlan, head of office for MP Graeme Downie, noted that staff are working beyond their paid hours to keep pace, prompting their office to halt responses to automated email campaigns. Lisa Gillmore of the GMB union confirmed that casework in several parliamentary offices has doubled or more.
In the private sector, AI-generated legal correspondence is worsening tribunal backlogs. Government figures reveal that between April and June, single-claimant employment tribunal filings jumped 28% year-over-year to 14,000 cases, bringing total open single claims to 70,000 by the end of June. Employee relations expert Jim Moore of Hamilton Nash observed that clients routinely receive overly verbose complaints that exaggerate minor incidents while omitting the specific resolution sought. One tech start-up founder reported expecting to pay £30,000 in legal fees to defend an employment tribunal case initiated by a former employee using AI-worded claims.
What it means for developers
As public and private organizations seek ways to handle incoming correspondence without losing human oversight, software engineers face a growing demand for intelligent document management, triage, and summarization infrastructure. Public bodies are already evaluating technological countermeasures; Wiltshire council is looking into AI tools to condense long emails and generate initial response drafts, while Stroud district council has discussed using Copilot for internal reporting.
For developers building solutions to combat AI clog, the challenge centers on designing pipelines that parse structured intent, filter out redundant fluff, flag hallucinations, and verify jurisdictional references. Building these systems requires testing multi-model architectures—comparing how different models summarize long documents or extract key actionable requests. Developers can try top AI models cheaply through one API at https://apixoai.online to build and benchmark these workflows efficiently.
Furthermore, developers building public-facing AI tools must account for output design. Systems that generate hundreds of pages of unchecked legal boilerplate without clear resolution criteria create friction for recipient APIs and human review teams alike.
Systemic responses and future governance
To prevent administrative collapse, experts and bodies are contemplating structural changes. Technology journalist Jamie Bartlett has emphasized that organizations must adopt automated filtering or establish stricter submission standards to avoid becoming entirely overwhelmed. Some ombudsman services have already implemented strict word caps and rigid form templates to limit attachment sizes and unstructured text.
Other proposed measures from researchers include implementing bot-blocking protocols on government web portals, introducing submission fees for certain requests, or instituting cognitive friction to slow down automated drafting. Meanwhile, official UK government guidance to civil servants stresses that while AI can assist internal workflows, human personnel must verify facts, safeguard personal data, and retain final decision-making authority.
Source: ‘AI clog’: UK public bodies battle onslaught of chatbot-generated complaints — The Guardian AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
