Why Britain’s Bureaucracy Can’t Rely On Friction To Curb AI Paperwork

🔍 Read the full analysis: Why Britain’s Bureaucracy Can’t Rely On Friction To Curb AI Paperwork on ThorstenMeyerAI.com

A Sky News report published October 20 found that written complaints, Freedom of Information requests and individual employment tribunal claims have risen across several UK public-service channels since ChatGPT launched in November 2022. The figures do not establish that AI caused the increases, but they sharpen a policy question for public bodies: whether to restrict submissions or redesign processes built around people writing letters and essays.

Using data reviewed by Sky News technology correspondent Rowland Manthorpe, the report describes a similar pattern across complaints about water companies, local authorities, GP practices, the Department for Work and Pensions and other services: volumes were broadly flat for years before beginning to rise around late 2022. The report says the growth continued and accelerated in many channels. Each service has its own circumstances, so the timing alone cannot show that chatbots caused the change.

The figures reported by Sky News vary by channel. Household complaints about water companies rose 144%; complaints about English local authorities, 78%; written complaints to GP practices, 71%; complaints to the DWP, 64%; and individual employment tribunal claims, 62%. Complaints about legal services rose 49%, written complaints to the NHS 35%, complaints about police forces 34%, and complaints about public services 23%. These are reported growth figures since ChatGPT launched; the material provided does not specify a separate measurement window or detailed methodology for each series.

Manthorpe also tested 100 recent FOI requests with an AI-text detector. The tool classified 89% as fully AI-written and 9% as human-written; a smaller 2019 sample was classified as entirely human. Manthorpe cautioned that AI detectors are unreliable and said the test does not prove AI caused the increase. Separately, a survey cited by Sky News found that 58% of civil servants surveyed said AI-generated material from the public had a significant impact on their work.

Access Versus Inbox Capacity

The issue is not simply whether AI makes it easier to send more correspondence. It is whether public services can respond to higher volumes without making it harder for people to challenge mistakes or claim benefits they are entitled to receive. Sky News reports that the proportion of DWP complaints upheld has fallen from the 50% range to close to 40%. That is a lower share, but the report says the larger total volume means the absolute number of upheld complaints is still higher than before.

That distinction matters when officials consider adding friction, such as fees, extra steps or limits on AI-written submissions. Such measures may reduce workload, but could also deter people who have a valid complaint and lack the time, confidence or writing skills to present it unaided. On the other hand, staff cited in the report describe a heavier, more complex inbox, and public bodies have finite capacity to assess each case fairly and promptly.

Former Government Digital Service founder Tom Loosemore argues that the deeper problem is reliance on written narratives as a gateway to services. His proposed direction is to collect structured information rather than ask people to write essays. That is an argument about service design, not a confirmed government plan. The report does not set out an approved programme, budget or timetable for such a change.

Guidance and Rising Claims

The Information Commissioner’s Office issued guidance in May after public authorities reported higher volumes of AI-generated FOI requests, including submissions that misquoted legislation or required clarification. The ICO’s position, as described in the Sky News report, is that AI-written requests remain valid. Authorities do not need to establish whether a person or an AI tool wrote a request, but may seek clarification when it is unclear or badly formed.

The pattern is not limited to Britain. A paper by researcher Chris Schmitz, reported by TechCrunch in September, identified 84 possible cases of what Schmitz calls “agentic flooding” across 11 jurisdictions. The report says many had been roughly flat before 2022 and increased afterward. It also cites the UK Housing Ombudsman, where complaints rose from about 2,600 in 2022 to just over 7,000 last year, and the US Consumer Financial Protection Bureau, where complaints grew fivefold. Those examples show wider pressure on complaint systems, but do not by themselves identify AI as the cause.

Some organisations have already used cost or procedural barriers. Sky News reports that the Financial Ombudsman Service found about a third of its complaints were AI-generated and began charging some professional representatives to bring complaints; complaint numbers then fell. That experience illustrates one way to reduce submissions, but the report does not establish whether the decline mainly removed weak cases, valid cases, or both.

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

— Civil servants quoted in the survey cited by Sky News

AI’s Role Is Not Proven

The report shows that several types of written submissions increased after ChatGPT’s launch, but it does not establish a causal link. Changes in service quality, public awareness, policy, population or complaint-handling practices may also affect volumes. The AI-detector test is a small exercise using a tool the report describes as unreliable; it cannot establish who wrote each request or why requests increased.

It is also unclear how much of the additional correspondence is fully automated, partly drafted with AI, or written without AI. Most submissions described in the report still involve people copying text between a chatbot and a form or letter. The scale and likely effects of AI agents that submit claims directly are not yet known. Nor does the source material provide a detailed breakdown of outcomes across all the complaint channels listed.

Service Design Under Pressure

The immediate next step for public bodies is to handle valid submissions under existing rules while deciding how to manage volume and complexity. The ICO guidance gives authorities a basis for asking for clarification on unclear FOI requests, but it does not authorize them to reject requests simply because AI may have been involved.

The wider question is whether departments and regulators will change their forms and case-management systems to collect structured facts, route cases and identify what a person is asking for without requiring a long written account. Sky News reports no specific national reform timetable or commitment. As AI agents make it easier to file submissions automatically, the pressure on complaint systems may grow; how quickly public services adapt, and whether proposed changes preserve access to redress, remains to be seen.

Key Questions

What did Sky News report?

It reported rising written complaints, FOI requests and individual employment tribunal claims across several UK channels since ChatGPT launched in November 2022. The figures show a timing pattern, not proof that AI caused the increases.

Does the report prove that AI caused more complaints?

No. The report documents increases and an AI-detector test on a sample of FOI requests, while noting that detectors are unreliable and the test is not conclusive. Other factors may also explain rising volumes.

Can public bodies reject a request because AI wrote it?

The ICO guidance described by Sky News says AI-written FOI requests remain valid. Authorities do not need to determine who or what wrote them, though they can ask for clarification when a request is unclear or badly formed.

What change does Tom Loosemore propose?

Loosemore argues for services that collect structured information instead of relying on essays and long written submissions. The report does not describe this as an adopted government policy or give a timetable for implementation.

Source: ThorstenMeyerAI.com