AI bots generated 76% of responses to UK vaping consultation opposing a disposable vape ban; DHSC excluded them
Researchers from the universities of Edinburgh and Bath found that AI bots generated 76% of responses to a UK government consultation on tobacco and vaping restrictions in October 2023, with 90,835 of 118,756 submissions machine-generated and almost all opposing a disposable vape ban. The Department of Health and Social Care detected the entries through automated bot-detection screening, after noticing high-frequency submissions from a small cluster of IP addresses and duplicated wording, and excluded them before assessing public opinion. Once they were removed, 79.6% of valid responses supported restrictions. The researchers did not establish who ran the campaign, and they warned that bodies with fewer resources could be vulnerable to similar manipulation.
What the AI did
Someone used AI bots to generate 90,835 of the 118,756 submissions to a UK government consultation on tobacco and vaping, almost all opposing a disposable vape ban. The submissions came at high frequency from a small cluster of internet addresses and used duplicated wording; who ran the campaign has not been established.
First reported October 27, 2026 · Added to the register November 4, 2026 · 2 sources
- Affected
- Department of Health and Social Care (DHSC)
- Country
- United Kingdom
- When it happened
- October 2023
- First reported
- October 27, 2026
What this means for you
Could this affect you?
If you run open online consultations or feedback forms. Government departments, local councils and other public bodies running open online consultations, especially those without specialist bot-detection screening, could be flooded the same way, as could organisations whose policy positions rely on consultation response volumes.
What to check
- Screen consultation and feedback submissions for bot activity before analysing results.
- Look for high-frequency entries from clustered internet addresses and duplicated wording.
- Combine several verification checks rather than relying on one.
- Publish how consultation responses are processed.
Areas of your AI programme this touches
Timeline
- October 2023Happened
- October 27, 2026First reported
- October 27, 2026A DHSC spokesperson said its controls against consultation manipulation worked effectively in this case and the fake responses were removed. Action on Smoking and Health called the evidence deeply concerning and said safeguards should be put in place, while the thinktank Demos described the findings as pretty alarming and said better screening and more transparency would help.[1]
Every fact and its source (9)
- Consultation periodOctober 2023“vaping restrictions in October 2023”[2]
- Machine-generated submissions90,835 of 118,756 submissions“Of 118,756 submissions, 90,835 were machine-generated”[2]
- Share of responses generated by AI bots76%“generated 76% of responses”[2]
- ResearchersUniversities of Edinburgh and Bath“researchers from the universities of Edinburgh and Bath”[2]
- Detection signalHigh-frequency submissions from a small cluster of IP addresses“sent at high frequency from a small cluster of internet protocol (IP) addresses”[2]
- Valid responses supporting restrictions79.6%“79.6% of valid responses supported restrictions on disposable vapes”[2]
- Campaign originNot established“They did not establish who was behind this operation.”[2]
- ResearcherDr Nason Maani, University of Edinburgh“Dr Nason Maani, from the University of Edinburgh”[1]
- Where disclosedResearch in the journal Tobacco Control“disclosed on Tuesday in research published in the journal Tobacco Control”[1]
Sources
- AI bots flooded 2023 UK government consultation on disposable vapes bantheguardian.com · October 27, 2026
- AI Bots Made Up 76% Of Responses In Failed Bid To Stop UK Disposable Vape Ban, But Officials Caught Themibtimes.co.uk · October 28, 2026
How this record is classified. Severity M (2/5): harm to individuals, or a policy breach with limited reach. OECD level: hazard, an event that could plausibly have led to harm. Evidence: Reported, meaning press reporting only.
The facts in At a glance and in Every fact and its source are quoted from the sources listed. The summary, What the AI did and What this means for you are written from those sources and checked against them automatically before publication. Records reflect what has been disclosed, not everything that has happened. Incidents are found through news feeds and the GDELT Project.
