The Unstoppable Flow Of AI Paperwork In Britain

🔍 Read the full analysis: The Unstoppable Flow Of AI Paperwork In Britain on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on monitors, keyboards and dev gear

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

A Sky News report published October 20 found that written complaints and claims across several British public-service channels have risen since ChatGPT launched in November 2022, while a survey cited in the report found that 58% of civil servants say AI-generated material from the public significantly affects their work. The figures show a broad increase, but do not prove that AI caused it or that the submissions are unfounded.

Technology correspondent Rowland Manthorpe reported growth in several channels, including household complaints about water companies, local authority complaints, GP practice correspondence, claims to employment tribunals and requests for information from public bodies. The reported increases since ChatGPT’s launch range from 23% for complaints about public services to 144% for household complaints about water companies. Complaints to English local authorities rose 78%, written complaints to GP practices 71%, and complaints to the Department for Work and Pensions 64%.

These figures describe changes in separate channels, each with its own circumstances; they should not be read as proof that AI caused every increase. Manthorpe’s report says the channels share a pattern of relatively stable numbers followed by rises from late 2022, but also notes that factors such as water companies’ recent difficulties may help explain particular trends. The data therefore show a timing and pattern worth examining, not a settled account of cause.

Manthorpe also tested 100 recent Freedom of Information requests with an AI-text detector. It labelled 89% fully AI-written and 9% human, while a smaller 2019 sample was labelled entirely human. The report cautions that AI detectors are unreliable and that this small test cannot establish who wrote the requests or whether AI drove the increase. Separately, a civil-servant survey cited by Sky News found that 58% of respondents said AI-generated public submissions had a significant impact on their work.

More Claims Reach Public-Service Inboxes

The reported increase matters because British public services rely on written applications, complaints and appeals to determine what people are entitled to receive. If AI makes it easier to produce formal correspondence, more people may be able to raise issues they previously struggled to set out. At the same time, staff may have to review more submissions, including longer or more complex ones, while continuing to distinguish valid claims from requests that need clarification.

The evidence cited in the report does not support treating volume alone as proof of abuse. At the DWP, the proportion of complaints upheld has fallen from the 50% range to close to 40%, according to Sky News. That is a lower success rate, but it remains substantial. Because the total number of complaints has grown, a lower share could still mean a greater number of upheld complaints in absolute terms; the report does not provide enough detail here to calculate that total.

The policy choice is consequential. Adding charges or hurdles might reduce submissions, but could also deter people with legitimate complaints. The report frames the alternative as changing public-service processes so they can handle more claims without relying on lengthy essays and letters.

The Shift From Letters to AI

For years, the number of Freedom of Information requests to British public bodies had changed little, according to the Sky News report. It says the line began to rise after ChatGPT’s November 2022 launch, with increases appearing across other written channels too. That sequence is the basis for the report’s argument that cheaper drafting tools may be changing how people interact with institutions, but chronology alone cannot establish causation.

In May, the Information Commissioner’s Office issued guidance on AI-generated FOI requests after public authorities reported higher volumes and more complex submissions, including some that misquoted legislation or required clarification. The ICO said such requests remain valid, authorities do not need to determine whether AI wrote them, and clarification can be sought when a request is unclear or badly formed.

The report also cites researcher Chris Schmitz’s account of 84 possible cases of “agentic flooding” across 11 jurisdictions. The term describes large-scale automated submissions. These cases are evidence of a wider concern, not proof that every rise in complaints is automated or artificial. The report says the UK Housing Ombudsman received about 2,600 complaints in 2022 and just over 7,000 last year.

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

— Civil servants quoted in a survey cited by Sky News

What the Rising Counts Cannot Prove

The Sky News figures establish reported increases across several channels, but the material provided does not give full underlying datasets, detailed calculation methods or consistent comparison periods for every measure. It also does not establish how much of each rise is attributable to AI rather than other developments, changes in awareness, service problems or shifts in access to complaint routes.

The detector test was small and, as the report acknowledges, AI-text detectors can be unreliable. A result labelled “AI-written” is not proof of authorship. It is also unclear how many public submissions are drafted wholly by AI, edited with AI assistance, or written without it. The 58% survey result reflects civil servants’ reported experience; the supplied material does not specify the survey’s sample size or fieldwork details.

Nor does the available information settle how many additional complaints are upheld in total, how staff workloads have changed in measurable terms, or whether the reported growth will continue. Those questions matter when deciding whether a policy response should focus on filtering submissions, changing forms, or adding staff capacity.

Redesigning Claims and Appeals

The immediate next step for public bodies is to apply existing rules to AI-assisted requests while managing unclear or complex submissions through clarification, as the ICO guidance describes. The source material does not identify a new nationwide policy or timetable for changing complaint and application systems.

Longer term, Loosemore argues that government should replace essay-style processes with services based on structured data. That approach could let people state what happened and what remedy they seek through clearer, more consistent fields, rather than making access depend on producing a polished letter. Whether departments adopt that approach, and how they would protect access for people with different needs, remains unsettled.

AI agents could add further pressure by submitting or pursuing claims automatically rather than merely helping someone draft a letter. The report raises examples such as parking appeals and rail delay compensation, but does not say these services have already been overwhelmed by automated agents. Public bodies will have to assess how such tools affect demand and service capacity without assuming that every additional claim is invalid.

Key Questions

What did the Sky News report find?

It reported rises since ChatGPT launched across several written channels, including complaints to public bodies and employment tribunal claims. The reported changes vary by channel, and the figures do not prove AI caused them.

Does the evidence show that most new complaints are false?

No. The report says the DWP’s share of complaints upheld has declined to close to 40%, but that remains a substantial proportion. Researcher Chris Schmitz is quoted as saying most cases he examined involved people claiming something they were entitled to.

Can public authorities reject an FOI request because AI wrote it?

According to the Information Commissioner’s Office guidance described in the report, AI-written requests remain valid. Authorities do not need to establish who or what wrote them, and can ask for clarification when a request is unclear.

How reliable was the AI-detector test?

It was a small test of 100 recent FOI requests, and the report warns that AI detectors are unreliable. Its results cannot establish authorship or prove that AI caused the broader rise in requests.

What changes could government make?

Tom Loosemore argues for services that collect structured information instead of relying on lengthy written applications. The report does not identify a government-wide plan or timetable for adopting that model.

Source: ThorstenMeyerAI.com