When the European Commission opened public consultation on revising two major tobacco laws—product regulations and advertising rules—it expected maybe a few hundred responses. Instead, over 80,000 flooded in within a month. That's 97 percent above the typical threshold for such consultations. What happened? An investigation by Dutch broadcaster Pointer reveals that Philip Morris International quietly deployed an AI-powered astroturfing tool to manufacture mass opposition, and Dutch citizens bought in wholesale: 71 percent of Netherlands submissions were machine-generated.

The 'Your Voice Your Choice' AstroTurf Engine

The scheme operated under the innocuous brand 'Your Voice Your Choice.' PMI recruited customers through QR codes plastered on posters at tobacco specialty shops and via their IQOS e-cigarette website. Users scanned the code and landed on a multi-step questionnaire asking about their smoking habits, product preferences, and which concerns mattered most to them. But here's the kicker: eleven of twelve selectable 'important aspects' argued against stricter regulation, while the single pro-regulation option was buried. The tool then generated personalized objections in first-person prose—ready to submit directly to the EU Commission under each user's name.

Matching AI Fingerprints at Scale

Researchers fed roughly 65,000 responses from fifteen EU countries through Pangram, an AI-detection system. After Portugal and France, Dutch submissions contained the highest proportion of machine-written content. To confirm origin, they ran statistical analysis—Jaccard similarity, TF-IDF, and Gaussian Mixture Models—comparing actual submissions against over 100 texts generated by PMI's own tool. The results were damning: 95 percent of responses flagged as AI-generated matched PMI's output so closely that attribution was near-certain. Of the 370 submissions traced to the campaign, only six advocated for stricter regulation. The rest were pre-poisoned with industry talking points about how restrictions 'backfire' or position vaping as a 'less harmful alternative.'

Experts Call It Straight-Up Manipulation

Rik Joosen, a public administration scholar at Leiden University specializing in EU lobbying, drew the line clearly: 'An AI-tool was effective here to get citizens to participate, but not with their fully own, sincere, and unguided voice.' Iskander De Bruycker, associate professor of EU lobbying at Maastricht University, went further: the branding 'Your Voice Your Choice' implies neutrality while Philip Morris only appears in fine print. 'From the perspective of democratic participation and ethical lobbying, this clearly crosses the line for me.' The European Commission itself expressed grave concern, stating that public consultations must reflect genuine opinion—not manufactured consent engineered by industry campaigns.

A 40 Percent Revenue Threat Fuels the Offensive

The timing isn't accidental. Approximately 40 percent of Philip Morris's revenue now comes from 'smoke-free' products—vapes, heated tobacco, and nicotine pouches—the very items facing stricter regulation under the proposed revisions. Under the WHO Framework Convention on Tobacco Control, which the EU has ratified, the industry should have zero role in policy formation due to inherent conflicts with public health. PMI's AI campaign represents a direct workaround: flood consultations with manufactured citizen voices to drown out legitimate public health concerns while maintaining plausible deniability.

Dutch Parliament Pushes Back

The exposure triggered political fallout. After doctors filed complaints, six parliamentary parties introduced a motion demanding the government condemn the campaign and push for its termination. Health Minister Sophie Hermans didn't oppose the motion but deferred enforcement to the Dutch Food and Consumer Product Safety Authority, which will assess potential violations of advertising law. Whether accountability follows remains uncertain—but the genie is out of the bottle.

Key Takeaways

  • Philip Morris deployed an AI tool branded 'Your Voice Your Choice' that generated 370+ fake citizen responses to EU tobacco consultations
  • 71 percent of Dutch submissions were AI-generated, with 95 percent matching PMI's tool output exactly
  • The questionnaire pre-loaded eleven anti-regulation arguments against a single pro-regulation option
  • WHO treaty prohibits tobacco industry involvement in policy-making—a line PMI explicitly crossed

The Bottom Line

This isn't clever marketing—it's the industrialization of democratic manipulation. PMI found a gap in EU consultation safeguards, weaponized generative AI to manufacture grassroots opposition at scale, and nearly got away with it. Regulators need to implement bot-detection requirements and verify that submitted comments actually reflect human deliberation—or these 'consultations' become nothing more than paid theater.