AI is helping people send complaints, freedom-of-information requests and consultation responses at a scale that officials say is straining Whitehall. The same tools are also being used inside government to process public input faster, leaving a live question about how to widen participation without drowning out people’s voices.
Watch Desk analysis
What happened
The Telegraph reports that civil servants are facing a surge in AI-assisted public submissions, including consultation responses and complex FOI requests. Government sources told the paper that AI agents can prepare responses on people’s behalf, while a sophisticated chatbot can assemble a 20-point FOI request in about 15 seconds. Cabinet Secretary Antonia Romeo described a “surge of automated, high-volume citizen demand” hitting operational desks.
The article cites University of Edinburgh researchers’ finding that bots submitted 76% of responses to a 2023 Department of Health consultation on disposable vapes. Officials identified 90,835 of 118,756 submissions as fraudulent, most opposing the proposed rules. That is a specific consultation, not a measure of how often AI-generated submissions affect government consultations overall.
Whitehall is also putting AI to work on the incoming material. Officials are using a “Consult” tool to speed analysis of more than 50,000 responses to a water-sector consultation. They believe it could eventually save 750,000 days of manual analysis a year. Separately, the article says Justice Transcribe has been used in 1.5 million meetings involving probation officers and is estimated to have saved 260,000 hours of civil service time.
Why it matters
Submitting a response has always been a way to take part in public decisions. AI may make that easier, but an avalanche of near-instant submissions can make it harder to distinguish considered views from automated volume. A consultation with tens of thousands of responses is not necessarily a better consultation if officials cannot interpret them, or if genuine voices get lost in the pile.
The answer is not simply to wave away anything written with AI. Tools that help officials analyse submissions could also help them handle more public input. The test is whether they make participation more meaningful, rather than turning the process into a contest between citizens’ keyboards and automation.
Our read
This is a useful warning about AI changing the workload of public services before any grand redesign of government arrives. But it is not a case for treating AI-assisted participation as inherently suspect: the reported bot finding concerns one consultation, and the same article describes tools intended to help officials manage the work. Whitehall needs to show how it separates coordinated or fraudulent volume from legitimate engagement, and whether its own tools make that judgement better.
What to watch
- Whether government publishes clearer evidence on AI-generated submissions across consultations.
- How officials distinguish fraudulent or coordinated responses from legitimate AI-assisted participation.
- Whether the projected savings from Consult translate into faster, more transparent analysis.
Discussion spark: Should public consultations accept AI-assisted responses on the same terms as any other submission, or should governments set limits when automated volume risks crowding out people’s views?
Sources and evidence
- Whitehall swamped by AI complaints – The Telegraph (15 November 2026, 15:00 UTC)
- Danny Boyle recreated Thatcher using AI without permission – The Telegraph (15 November 2026, 15:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.