What happened
The Guardian has reported that UK public bodies are struggling with a surge of correspondence written with AI chatbots. Councils say complaints that used to run to around two pages are now often ten times longer, with one example reaching about 200 pages. Transport for London received a record 4,769 freedom of information requests in 2025-26 and believes a growing share of the appeals it handles are AI-generated. MPs’ offices describe the same pattern. One senior caseworker told the paper that half of the emails her office receives involve AI in some form.
The effects reach well beyond complaints. Councils cite objections to local planning applications among the AI-assisted correspondence they now receive, alongside parking appeals and social care complaints. Some ombudsman services are responding by introducing word limits and online templates. Researchers have even coined a term for the trend: ‘agentic flooding’. And it’s not all one-sided. Government is adopting AI too: its GOV.UK Chat assistant now averages more than 7,000 users a month, according to a recent parliamentary answer.
Why it matters for communicators
AI has removed most of the effort from writing a detailed, formal response. In one sense that is a good thing. People who once felt they lacked the confidence or vocabulary to challenge a decision can now put their case fluently. That widens access, and no organisation should want to shut it down or treat AI-assisted views as less valid.
But it also changes what a consultation inbox tells you. For years, the number and length of responses has been treated as a rough measure of how strongly a community feels. That measure is breaking. A large volume of long, well-argued submissions might reflect deep local concern. It might equally reflect a small number of people using the same tool to produce polished text at speed. At a glance, the two can look identical.
For those in development, infrastructure, housing and public bodies, the risk runs in two directions. First, sheer volume can slow decisions, stretch officer time and add cost to programmes that are already under pressure. Second, and more seriously, genuine community voices can get lost in the noise. The resident who turns up to an event with one clear concern about traffic outside a school matters just as much as a 200-page submission and is likely to be more representative.
There is a reputational dimension too. If people feel their views were filtered out or dismissed because of how they were written, trust suffers, and so does the legitimacy of the final decision. How an organisation handles AI-assisted responses will increasingly become part of the story about whether a consultation was fair.
What this means for organisations
- Design for quality, not just quantity. Structured questions, clear response formats and plain-English materials make submissions easier to analyse consistently and fairly.
- Meet people in person. Drop-in events, workshops, door-knocking and conversations with community groups give a truer picture than inbox volume alone, and will reach people who would never write in.
- Plan your capacity and be transparent. Budget time and resource for longer, more complex responses, and explain clearly how every view, however it might be written, has been considered.
How JBP can help
JBP runs consultation and engagement for projects across the built environment, infrastructure, housing and transport. We help clients design processes that are open, robust and focused on hearing the real voices in a community, from early stakeholder mapping to face-to-face events and clear reporting. To discuss your next consultation, get in touch with our team.