Artificial Intelligence for Addiction: Advancing Prevention, Treatment, Recovery and Crime Mitigation

Substance-use disorders remain a complex public-health and social challenge, affecting individuals, families and communities while contributing to preventable illness, social exclusion, drug-related crime and pressure on health, justice and social-care systems. Artificial intelligence (AI) offers new opportunities to strengthen responses across the addiction continuum by combining clinical, behavioural, social and public-health data to support earlier prevention, more personalised treatment, sustained recovery and evidence-informed crime mitigation. Emerging applications include risk screening, early-warning systems, digital therapeutics, conversational agents, treatment-retention prediction, relapse-prevention support, resource referral and the analysis of drug-market trends. A systematic review of chatbot interventions found that most evaluated programmes focused on treatment, while a smaller proportion addressed prevention and assessment, indicating both growing evidence and important gaps in the field.

This article examines the potential contribution of AI across four interconnected domains. First, AI can support targeted prevention by identifying patterns associated with vulnerability, improving population-level surveillance and enabling timely, culturally appropriate interventions. Second, predictive and adaptive systems may assist clinicians in screening, treatment planning, care coordination and the identification of patients at risk of disengagement or relapse. Third, AI-enabled mobile applications, chatbots and monitoring tools can provide continuous recovery support between clinical appointments through reminders, craving-management strategies, self-monitoring and referrals to human services. However, such tools should complement rather than replace professionals, peer workers and therapeutic relationships; early pilot research highlights the importance of human oversight, escalation procedures and accurate information.

Fourth, AI may contribute to crime mitigation by improving the analysis of drug-trafficking trends, identifying emerging threats and supporting the allocation of prevention and enforcement resources. For example, UNODC describes a near-real-time, multisource platform intended to improve early warning regarding drug threats and trafficking patterns. Nevertheless, the use of AI in policing and criminal justice creates substantial risks. Biased historical data, opaque decision-making and feedback loops can reproduce discrimination, intensify surveillance and contribute to over-policing of already marginalised communities. In the European Union, several law-enforcement applications—including systems assessing the risk of offending or reoffending—fall within the high-risk framework of the EU Artificial Intelligence.

The article argues that AI should be understood as a decision-support and public-health infrastructure rather than an autonomous solution to addiction or crime. Responsible implementation requires representative and high-quality data, privacy protection, informed and dynamic consent, explainability, independent evaluation, continuous monitoring, equitable access and meaningful participation by people with lived experience. The World Health Organization’s principles—protecting autonomy, promoting safety and public benefit, ensuring transparency, accountability, inclusiveness and sustainability—provide a relevant ethical foundation. A human-centred, rights-based and interdisciplinary approach can enable AI to improve prevention, treatment and recovery while avoiding the criminalisation, stigmatisation and exclusion of people who use drugs.