British government agencies, local councils and the court system are facing a sharp rise in submissions, many of them drafted with the help of AI chatbots. Complaints are becoming longer, containing repetition, irrelevant arguments and legal language of questionable applicability. As a result, staff are spending more time processing documents, putting additional pressure on the budgets of already overstretched bodies.

In the UK, the problem is already being called “AI clog” — the clogging of government and legal procedures with AI-generated text, TechRadar reports.

More complaints, harder to process

Before AI chatbots became widely available, preparing a substantial complaint took time: people had to investigate the circumstances of their case, find relevant regulations and formulate their arguments themselves. This created a natural barrier that filtered out some unfounded or poorly prepared submissions.

Now, all it takes is to describe the situation to ChatGPT or another chatbot and ask it to draft a persuasive complaint. The resulting text can be sent almost immediately to a government agency, local council, court or company.

For complainants, this makes it considerably easier to access complaints and appeals procedures. But on the receiving end, each submission still has to be reviewed by a person. Staff cannot simply ignore a lengthy document: they need to understand the substance of the complaint, check the circumstances and determine whether it requires a response or further action.

Holly Wickens, a senior caseworker in a Labour MP’s office, told The Guardian that roughly half the emails they receive are prepared with AI assistance to some degree. According to her, these submissions are considerably harder to deal with.

Hundreds of pages of legal text — and AI errors

The problem lies not only in the number of documents but also in their content. Staff at public bodies encounter very long, repetitive texts containing extraneous reasoning and demands unrelated to the matter under consideration.

Some submissions can run to hundreds of pages. One case described by The Guardian involved a document of around 200 pages.

Yet an impressive length and a confident legal style are no guarantee of sound arguments. AI can invent facts, misinterpret legislation, cite nonexistent provisions or reference real sources that have no bearing on the particular case.

As a result, specialists have to do more than examine the complaint: they must separate the actual facts from AI-generated reasoning, verify references and determine which demands genuinely have a basis.

The consequences are particularly evident in the courts. The UK already faces substantial delays in hearing cases. Now companies, including small businesses, are having to spend additional money on proceedings arising from complaints by former employees who use AI to prepare legal documents.

One business owner who faced this situation summed it up as follows: “Now everyone has a free lawyer.”

How authorities are trying to cope with the influx of complaints

Government agencies and companies are considering several ways to reduce the workload. One option is to limit the amount of text that can be submitted through a complaints form. Another is to replace free-text fields with standardised templates requiring complainants to specify the circumstances, their demands and supporting evidence.

Such measures could make it harder to submit lengthy, repetitive texts, but they also call for caution: overly strict limits could prevent people with legitimate complaints from explaining their situations in detail.

Another option is to use AI to process incoming documents. Some British local councils have already introduced tools that help summarise information, prepare documents and draft emails. This allows staff to get to grips with the material and prepare responses more quickly.

However, automating complaint processing creates a kind of feedback loop: AI helps complainants produce more documents, and now the same technology is being proposed as a way to process that influx. Meanwhile, errors and inaccuracies can arise on both sides of the process.

Access to justice or a new bureaucratic problem?

The spread of AI also has an obvious upside. People who lack legal knowledge and cannot afford a professional adviser now find it easier to formulate a complaint, understand their rights and approach the appropriate authority.

But lowering the barrier to filing complaints does not automatically improve their quality. When preparing a lengthy legal document costs virtually nothing, complainants may be tempted to submit texts without checking every reference or assessing how well the arguments fit their particular situation.

Ultimately, the time saved by complainants translates into additional work for staff, who have to check every submission. AI has indeed made legal assistance more accessible, but it has also increased the cost of processing complaints for those obliged to review them.

The main challenge now is to preserve the benefits of AI’s accessibility without turning government and court procedures into an endless exercise in checking automatically generated documents. Simply introducing new tools will not be enough: users also need to understand that a persuasively written text does not necessarily make a legally sound complaint.