Artificial intelligence: an emerging lever to strengthen tobacco control on a global scale

As tobacco and nicotine consumption patterns evolve rapidly, tobacco control stakeholders are increasingly interested in artificial intelligence (AI) as a tool to support public health policies. Monitoring violations, detecting illegal marketing practices, and supporting smoking cessation: AI has numerous applications and is attracting growing interest in order to better anticipate, regulate, and counter the persistent influence of the tobacco industry.

Growing recognition of the role of AI at the Dublin conference

The World Conference on Tobacco Control, held in Dublin from June 23 to 25, 2025, highlighted the growing interest in new technologies, particularly artificial intelligence, as a strategic lever to strengthen tobacco control. For the first time, several sessions were explicitly dedicated to the use of AI to monitor the industry's digital marketing strategies, detect illicit content online, and support smoking cessation policies.[1].

This development reflects a shared awareness of the limitations of traditional approaches in the face of the tobacco industry's rapid adaptation, particularly in targeting young people via social media. Through various workshops and feedback sessions, participants highlighted the potential of AI to fill regulatory gaps, provide actionable data to health authorities, and build capacity in low-resource countries.

The mobilization of young actors, particularly through initiatives such as Global Youth Voices, also illustrated the growing appropriation of these tools for the purposes of advocacy, citizen monitoring and social justice.

Promising technology to strengthen surveillance, particularly in low-income countries

In many low- and middle-income countries, institutional and human capacity remains limited to ensure continuous and effective monitoring of violations related to the promotion of tobacco and nicotine products. In this context, artificial intelligence could be a particularly useful tool to strengthen the enforcement of tobacco control policies.

Thanks to its automated analysis capabilities, AI can process large volumes of data from social media, video platforms, and e-commerce to detect in real time content promoting nicotine products, often distributed covertly or circumventing national legislation. It thus offers a way to more quickly identify violations, map tobacco industry circumvention strategies, and document marketing campaigns targeting young people.

In countries where human resources dedicated to regulation are insufficient, the integration of artificial intelligence tools could considerably lighten the workload of health authorities and support civil society organizations in their monitoring and alert activities. Experiments conducted in Africa and Asia show that AI can facilitate the identification of violations, produce automated reports, and fuel awareness-raising or advocacy actions based on tangible data.[2].

Beyond simple detection, artificial intelligence can also help us better understand emerging narratives surrounding nicotine products, including in local languages or on platforms that are difficult to access with traditional monitoring systems. It thus represents a potential lever for strengthening fairness in global tobacco industry surveillance, provided there is equitable access to tools, appropriate technical support, and rigorous ethical oversight.

A potential to structure and supervise

While artificial intelligence offers promising prospects for strengthening prevention, monitoring, and regulation in the field of tobacco control, its use also raises significant challenges that must be anticipated.

First, the issue of ethics and data governance is central. AI systems rely on the collection and processing of sometimes sensitive personal data, particularly in the context of weaning or behavioral monitoring applications. It is therefore essential to ensure a robust protection framework, compliant with international standards on confidentiality, algorithm transparency, and informed user consent.

Furthermore, several experts warn of the risks of algorithmic bias, which could reproduce or amplify certain existing inequalities. Particular attention must be paid to the quality, diversity, and representativeness of the data used to prevent already vulnerable groups from being excluded or poorly targeted by the tools developed.

Another point of vigilance: the potential appropriation of these technologies by the tobacco industry itself, particularly through initiatives presenting themselves as innovative or responsible. Without clear oversight, some players could use AI to improve their image, influence anti-smoking measures, or circumvent control mechanisms by adapting their advertising content to the limitations of detection systems.

Finally, the growing environmental impact of AI deployment cannot be ignored. Training and operating artificial intelligence models, particularly large language models or visual recognition systems, requires significant computing power and consumes a significant amount of resources (water and energy). This ecological cost raises questions of sustainability, particularly in the context of environmental transition and the search for exemplary public health policies.

For these reasons, the mobilization of artificial intelligence in tobacco control must be part of a coherent, structured strategy based on the principles of transparency, fairness, accountability, and sustainability, experts emphasize. It can only produce concrete and acceptable results through close cooperation between researchers, public authorities, civil society actors, and international institutions, ensuring that the tobacco industry is excluded.

©Generation Without Tobacco

AE

[1] John Musenze, AI, the new frontier for global tobacco control, Scidev, published June 25, 2025, accessed June 27, 2027

[2] David B. Olawade, Charity A. Aienobe-Asekharen, Artificial intelligence in tobacco control: A systematic scoping review of applications, challenges, and ethical implications, International Journal of Medical Informatics, Volume 202, 2025, 105987, ISSN 1386-5056, https://doi.org/10.1016/j.ijmedinf.2025.105987.