Three-quarters of the responses to the UK government’s 2023 consultation on creating a smoke-free generation and tackling youth vaping were submitted by bots, according to a peer-reviewed study published in the journal Tobacco Control on October 27, 2026. The Department of Health and Social Care identified 90,835 of the 118,756 responses as fraudulent using bot detection methods, and excluded them from its decision-making, reversing the apparent direction of public opinion on disposable vapes.
A Flood Aimed At One Question
The bot responses were strikingly uniform. All of them consisted solely of multiple-choice selections, with no free-text answers, and 90,833 of them answered only three questions, with more than 99.9% disagreeing with proposed restrictions on disposable vape sales. The same number reported that they had learnt of the consultation via social media, a pattern the researchers say points to a coordinated campaign rather than organic public engagement.
The Department of Health and Social Care said the fraudulent responses were identified where the data submitted was incompatible with the survey flow followed by genuine respondents who completed the form manually. Its consultation report noted that such responses were often duplicated many times and sent at high frequency from a small number of IP addresses, which the researchers say is consistent with a mass, automated campaign to influence the consultation.
Had the bot responses been counted as genuine, 81.3% of respondents would have appeared to disagree with restrictions on disposable vape sales. Once they were removed, leaving 27,921 valid responses, 79.6% of valid respondents supported the restrictions, the reverse of what the unfiltered dataset would have indicated. The attempt did not change the outcome, because the department caught and excluded the submissions, but it came close to manufacturing a public mandate that did not exist.
The First Documented Case Of Its Kind
The researchers, Nason Maani of the University of Edinburgh and the London School of Hygiene and Tropical Medicine and Anna Gilmore of the University of Bath, describe the episode as the first documented instance of large-scale automated interference in a formal public health policy consultation. Bots have previously been used to spread health misinformation on social media, they note, but this is the first peer-reviewed evidence of them being deployed at scale against a government consultation itself.
The study is careful about what it cannot prove. The researchers could not independently verify the department’s bot detection methods, because the details are not publicly available and respondent metadata was withheld under freedom of information exemptions. It is also impossible, on the information received, to attribute the source, intent or commercial sponsorship of the campaign.
Still, the authors draw a pointed parallel. While it is not possible to ascertain the commercial origins of this campaign, they write, transnational tobacco companies have previously sought to swamp public consultations with favourable responses, using mass recruitment campaigns and postcard submissions. The scale and directional consistency of the bot responses, they argue, are consistent with prior evidence of tobacco industry efforts to manipulate public consultations through mass coordinated submissions.
Why The Safeguards Held, And Why They May Not Next Time
The UK’s safeguards worked in this case: the department’s bot detection caught the submissions, excluded them from the consultation report, and the legislation proceeded. The Tobacco and Vapes Act received royal assent on April 29, 2026, making it illegal to ever sell tobacco products to anyone born on or after January 1, 2009, and restricting the supply and sale of disposable vapes. The consultation’s reported result, based on valid responses, supported the restrictions the bots had opposed.
The warning in the study is about the next campaign, not this one. As bots grow more sophisticated, future attempts may be harder to detect, producing varied, plausible answers from many machines rather than duplicated selections from a handful of IP addresses. Consultations in settings with fewer resources or weaker screening tools would be at greater risk, and campaigns often take place with little or no public awareness.
The authors call for stronger detection capacity and governance safeguards, including IP-based rate limiting, CAPTCHA-style checks, explicit conflict-of-interest screening questions and wider cyber security capacity to accompany increases in correspondence. Technical barriers alone, they note, only raise the bar; the incentive to influence consultations remains, and every government that asks for comments online faces the same risk.
Why It Matters
This matters far beyond vaping. Public consultations are how governments claim democratic legitimacy for regulation, and if their results can be flooded at scale, that legitimacy becomes cheap to fake, a risk that runs alongside the wider debate over AI and public trust captured in recent AI policy controversies. The UK case shows both sides of the lesson: detection and exclusion can protect a decision, but only if the screening exists in the first place, and only while bots remain clumsy enough to catch, much as governments are already learning with AI adoption across public agencies.
Background
The UK government consulted in 2023 on creating a smoke-free generation and tackling youth vaping, publishing its response in early 2024 after excluding 90,835 bot-identified responses. The resulting Tobacco and Vapes Act received royal assent in April 2026, and the new study is the first peer-reviewed examination of the automated submissions behind the consultation.
Source: Tobacco Control study and study summary
0 Comments