Britain’s Bureaucracy Meets The AI Paperwork Surge

🔍 Read the full analysis: Britain’s Bureaucracy Meets The AI Paperwork Surge on ThorstenMeyerAI.com

A Sky News report published October 20 found that written complaints, Freedom of Information requests and employment tribunal claims have increased across several UK public-service channels since ChatGPT launched in November 2022. The figures show a broad rise, not proof that AI caused it; the report also cites civil servants who say AI-generated material is adding substantially to their workload.

Using data seen by Sky News, technology correspondent Rowland Manthorpe compared activity in multiple channels since ChatGPT’s launch. Household complaints about water companies rose 144%, while complaints about English local authorities increased 78% and written complaints to GP practices rose 71%. The reported figures also include increases of 64% for complaints to the Department for Work and Pensions (DWP), 62% for individual employment tribunal claims and 49% for complaints about legal services.

Sky News reported smaller rises in written complaints to the NHS, at 35%; complaints about police forces, 34%; and complaints about public services, 23%. The figures are comparisons since ChatGPT launched; the source material does not specify a uniform measurement window or establish how much of each increase is attributable to AI. Each channel has other possible influences. Water companies, for example, have faced difficult years, the report says.

Manthorpe also tested 100 recent FOI requests with an AI-text detector. It classified 89% as fully AI-written and 9% as human-written; a smaller sample from 2019 was classified as entirely human. The report cautions that detectors are unreliable and that this test cannot prove AI caused the wider rise. A civil-servant survey cited by Sky News found 58% said AI-generated content from the public significantly affects their work.

More Claims, More Pressure on Services

The rise matters because many public bodies handle applications, complaints and appeals through written submissions. If AI makes it quicker and easier to draft those submissions, authorities may have to process more material, including requests that are incomplete, legally inaccurate or difficult to assess. That can add work for staff and slow responses for other people seeking help.

But a higher volume cannot automatically be treated as a surge in abuse. Researcher Chris Schmitz, whose study of possible “agentic flooding” cases was reported by TechCrunch in September, said most cases he reviewed involved people seeking something they were entitled to claim. At the DWP, Sky News reported, the proportion of complaints upheld has declined from the 50s to close to 40%. Because the total number of complaints rose, that lower share does not by itself show that fewer complaints are justified in absolute numbers.

The policy choice affects access as well as workload. Fees or other barriers could reduce submissions, but could also deter people with valid complaints. The question for government is how to handle more correspondence without making it harder for people to challenge decisions or claim what they are owed.

The FOI Rules and Earlier Signals

The increase is not confined to one UK institution. A paper by Chris Schmitz, covered by TechCrunch in September, documented 84 possible cases of “agentic flooding” across 11 jurisdictions. The source describes many as broadly flat before 2022 and rising afterward. The UK Housing Ombudsman received about 2,600 complaints in 2022 and just over 7,000 last year, while complaints to the US Consumer Financial Protection Bureau grew fivefold, according to the supplied report.

In May, the Information Commissioner’s Office (ICO) issued guidance on AI-generated FOI requests after public authorities reported higher volumes and more complex submissions. Some requests, they said, misquoted legislation or needed substantial clarification. The ICO’s position, as described in the report, is that an AI-written request remains valid: authorities do not need to determine who or what wrote it, but may seek clarification if it is poorly formed.

The Financial Ombudsman Service has taken a different practical step, the report says. After finding that about a third of its complaints were AI-generated, it began charging some professional representatives to bring complaints; the number subsequently fell. That example shows that rules can affect volume, but it does not establish whether the reduced complaints were invalid or whether the approach would suit other public bodies.

“It’s like being bombarded with correspondence at speed.”

— Civil servants responding to an academic survey cited by Sky News

How Much AI Drives the Increase

The reported trend does not isolate AI as the cause. The figures cover different services and complaint channels, and the source material does not give a common baseline definition or full methodology for every percentage. Other events and service-specific problems may have contributed. The detector test is also not conclusive: AI-text detectors can be unreliable, and the sample of 100 recent FOI requests cannot establish how representative the results are.

It is also unclear how much added volume consists of valid claims, duplicate or poorly formed submissions, or automated activity. The available figures on upheld DWP complaints offer one partial indicator, but the reported change in the proportion does not settle the question across all departments and regulators. Nor does the Financial Ombudsman example show whether its policy reduced only unsuitable complaints.

The report describes a coming shift toward AI agents that can submit claims for users, rather than merely help draft them. The scale and timing of that shift, and its effect on public services, remain uncertain. Sky News demonstrated an agent service from Meta, but the supplied material does not show that automated agents are already responsible for the reported increases.

Redesigning Applications and Appeals

The immediate question for public authorities is how to handle requests that are valid but arrive in greater numbers or need clarification. The ICO guidance gives one current reference point for FOI teams: assess the request rather than trying to identify its author, and ask for clarification when needed. The supplied reporting does not identify a new government-wide policy or timetable for responding to the wider rise.

Loosemore’s proposed direction is to replace essay-heavy processes with structured data and clearer forms. He has warned that adding friction could prompt technology providers to find ways around it, while also excluding people who need help expressing a legitimate claim. His view is a proposal, not an announced government plan.

Further reporting will need to establish whether the upward patterns continue, how agencies distinguish valid claims from incomplete or automated submissions, and whether service redesign can reduce administrative work without restricting access. For now, the confirmed picture is a multi-channel rise reported by Sky News and a substantial share of surveyed civil servants reporting work effects—not a proven, single-cause AI surge.

Key Questions

What did the Sky News report find?

It reported increases since ChatGPT launched in November 2022 across several UK channels, including water-company complaints, local-authority complaints, FOI requests and employment tribunal claims. The percentages vary by channel and do not prove AI caused the increases.

Does the report prove AI caused more complaints?

No. The report describes a timing pattern and results from a small AI-detector test, while warning that detectors are unreliable. Other factors may also explain increases in individual services.

Are AI-written FOI requests valid?

According to the ICO guidance described by Sky News, an AI-written request remains valid. Authorities do not need to establish who or what wrote it and can ask for clarification if the request is unclear or badly formed.

Are the extra complaints mostly fraudulent?

The supplied evidence does not support a conclusion that the increase is mostly fraudulent. Researcher Chris Schmitz said most cases in his study involved people seeking something they were entitled to claim, though the findings do not classify every UK submission.

What changes could public services make?

Tom Loosemore has argued for services that use structured information rather than relying on long written applications. No government-wide redesign plan or timetable is identified in the supplied report.

Source: ThorstenMeyerAI.com