Study: bots produced about three-quarters of responses to a 2023 UK consultation on vapes

A study in the journal Tobacco Control reports that most of the responses submitted to a 2023 UK government consultation on smoking and vaping did not come from real people. Researchers at the University of Edinburgh found that roughly three-quarters of the submissions were identified as automated bot responses, according to UK media reports published on October 27.

What the researchers found

The consultation, run by the Department of Health and Social Care (DHSC), covered plans for a smoke-free generation and measures against youth vaping, including restrictions on disposable vape sales. According to GOV.UK, it was open from October 12 to December 6, 2023, and the government published its response in January 2024.

The researchers analysed published consultation reports and data obtained from DHSC through Freedom of Information requests. Those records show 118,756 responses, of which 76%, more than 90,000, were classed as fraudulent by the department's bot-detection methods and excluded.

According to the reports, almost all of the bot responses answered multiple-choice questions in opposition to restricting disposable vape sales, and the same share said they had heard about the consultation through social media. Once they were removed, 79.6% of valid responses backed the disposable vape proposals. The authors describe the episode as the first documented case of large-scale automated interference in a formal public health policy consultation, and say its scale and consistency point to a coordinated campaign.

What remains unclear

Headlines describe the submissions as generated by "AI bots", but the coverage reviewed here does not say whether large language models or other generative AI were involved; the study's findings as reported refer to bots and automated responses. The researchers also say it is not possible to determine the commercial origin of the campaign. They note that tobacco companies have previously tried to sway consultations through mass recruitment and postcard campaigns, without linking this case to any company.

The authors warn that detection methods need to keep pace as bots become more sophisticated, and that consultations with fewer resources for screening could be more exposed.

Government response

DHSC said its controls worked in this case, that the fraudulent responses were removed and that the consultation still informed its smoke-free generation policy.