1. Introduction
The illicit tobacco trade (ITT) remains a global threat, undermining health policies and fiscal sustainability. Many countries, including Malaysia, have increased taxes on cigarettes and have adopted more stringent tobacco control policies to combat smoking-related morbidity and mortality. Yet these efforts have fuelled the production and consumption of smuggled and illegal tobacco products, making illicit cigarette trafficking a growing global concern (Paraje et al., 2022). Regulatory shortcomings allow criminal organisations to bypass taxation, thereby causing economic losses while also endangering consumers, who are exposed to unregulated and often more harmful products (Bialous, 2016). Addressing this growing issue requires innovative approaches, and artificial intelligence (AI) has emerged as a promising tool to enhance enforcement, strengthen illicit supply chain detection and support public health protection (Bernama, 2024; World Health Organization (WHO), 2023). Building on this, the present study examines the role of AI in strengthening enforcement in Malaysia.
The ITT accounted for 11.6 per cent of global cigarette sales, or 650 billion illicit cigarettes, marketed in 2007. This resulted in a global tax revenue loss of USD40.5 billion (Table 1; Joossens et al., 2010). Notably, a large proportion of this trade, estimated at 533 billion cigarettes, originates from low- and middle-income countries (LMICs), which lose approximately USD22.9 billion in tax revenue annually. WHO (2023) emphasises the need to eradicate all forms of illicit cigarette trade, including large- and small-scale smuggling, counterfeit products and illegal manufacturing. These factors make effective tobacco control more challenging. Article 15 of the WHO Framework Convention on Tobacco Control (WHO, 2003) provides guidance on how parties can address and eliminate illicit trade activities. It also emphasises tracking and tracing systems, cross-border monitoring and stronger enforcement measures to combat ITT (WHO, 2003). These measures are designed to make illicit activities more difficult to sustain. These global estimates provide context for understanding the scale of the problem addressed in this study.
Building on this, the Parties to the WHO Framework Convention on Tobacco Control established the Protocol to eliminate illicit trade in tobacco products (WHO, 2013), which entered into force on 25 September 2018. This protocol is a comprehensive package of measures against illicit trade. It seeks to halt illegal trade, strengthen enforcement and foster cross-border cooperation. These measures support cross-border monitoring, cargo tracing and coordinated customs enforcement activities. In this context, AI-enabled enforcement approaches may further strengthen regulatory monitoring, cross-border surveillance and illicit tobacco detection mechanisms.
1.1. Illicit tobacco trade in Malaysia
Illicit tobacco is a global challenge, and Malaysia is significantly affected by ITT. Without independent data on this issue, much of the available information derives from studies commissioned by British American Tobacco (BAT) and Japan Tobacco International (JTI), who are major contributors to illicit tobacco market reporting (BAT, 2020; Nielsen Consumer LLC, 2024; Oxford Economics, 2019). One report asserts that approximately MYR4.5 billion (USD1.1 billion) in excise duties to the Malaysian government was lost in 2023 and that illicit trade accounts for 55.6 per cent of the total tobacco market (Nielsen Consumer LLC, 2024). The Oxford Economics report (2019) ranks Malaysia as the world’s top consumer of illegal tobacco by market share (Table 1). Much of this evidence is derived from industry-supported reports, highlighting the need for independent analysis, particularly in evaluating enforcement-related challenges and AI-based intervention strategies. These enforcement challenges have increased interest in AI-assisted customs surveillance, predictive risk profiling and automated detection systems to strengthen illicit tobacco monitoring in Malaysia.
Tobacco is used by around 19 per cent of the adult population in Malaysia. Malaysia’s strategic location in Southeast Asia increases its vulnerability to illicit tobacco smuggling. The major ports, including Johor (17%), Selangor (15%), Pahang (13%), Sabah (11%) and Sarawak (10%), account for a total of 66 per cent of the nation’s trans-shipment activities, making them among the busiest in the country (Table 1; Nielsen Consumer LLC, 2024). Malaysia is also surrounded by neighbouring countries, including Indonesia, Thailand, Singapore, Brunei and the Philippines, that provide further opportunities for tobacco smuggling via sea and land (Rejab & Zain, 2006; Tiigah & Siu, 2020). Illicit tobacco is smuggled through various methods, including false declarations during trans-shipment or directly, such as kretek transported by small boats and distributed via trucks (Tam et al., 2018). Moreover, illicit locally produced tobacco products are frequently sold with counterfeit tax stamps, widely accessible in shops and street markets (Tam et al., 2018; The Star, 2012). These patterns highlight the complexity of illicit tobacco distribution networks and the need for more effective monitoring and enforcement mechanisms. These evolving smuggling methods complicate conventional customs enforcement, particularly shipment screening, cargo monitoring and border inspections.
Cigarette consumption in Malaysia, both legal and illicit, was estimated at 19.1 billion in 2017 (Table 1). Total illicit cigarette consumption in Malaysia increased from 2016 to 10.6 billion in 2017, representing a 10.7 per cent increase from 2016. This increase was mainly driven by the growing supply of contraband white cigarettes. In 2017, the estimated tax loss from illicit cigarette consumption in Malaysia was MYR4.7 billion (Table 1). These trends demonstrate the continuing economic and enforcement challenges associated with ITT in Malaysia. The overall illicit cigarette incidence in Malaysia in 2024 was reported at 55.6 per cent (Nielsen Consumer LLC, 2024), reflecting the ongoing challenges in addressing ITT. This includes 8.2 per cent of products featuring counterfeit tax stamps, 39.5 per cent of smuggled whites, and 7.4 per cent of smuggled kretek (Table 1). Customs authorities face increasing pressure to strengthen shipment monitoring and intelligence-led enforcement operations.
Vijayan and Chethiyar (2020) and Vijayan et al. (2022) highlight the broader impacts of the ITT. The ITT presents complex challenges affecting Malaysia’s economic security and national safety. It results in a loss of revenue and tax contributions, which have significant negative implications for public goods such as health and education. Furthermore, the ITT supports organised crime, posing a threat to national security and leading to other illicit activities. The government strengthened penalties, including imprisonment and fines of MYR100,000 for selling illicit cigarettes and approved amendments to the Customs Act 1967 and Excise Act 1976 (Royal Malaysian Customs Department, 1976), in January 2019. Conventional enforcement measures, such as manual data collection, physical checks and border patrols, are resource-intensive and insufficient to counter advanced tactics used by these networks (Pahlevi, 2024). AI-driven customs intelligence systems may improve surveillance efficiency, anomaly detection and shipment risk profiling. This study therefore examines how AI can enhance enforcement effectiveness in the Malaysian context.
Recent studies reported that AI is increasingly applied in enforcement contexts to analyse large datasets, detect smuggling trends and automate surveillance (Kempen, 2024; Mademlis et al., 2025). Predictive analytics, for example, is used to estimate high-risk zones and timings for smuggling activities, leading to targeted and efficient resource allocation. For example, machine learning (ML) algorithms scan data from customs declarations, shipping manifests and surveillance footage to detect anomalies associated with the ITT. AI-based technologies such as facial recognition and licence plate recognition systems enhance monitoring at borders and checkpoints, reducing human error and improving operational efficiency. Natural language processing (NLP) technologies also monitor online marketplaces, social media and the dark web for suspicious activity and tracking keywords and coded language used in illegal sales (Barrera et al., 2019). NLP technologies also analyse customs records, financial transactions and legal documents to detect fraud, identify counterfeit products and assist law enforcement. AI-assisted surveillance and anomaly detection systems may improve customs targeting efficiency and enforcement coordination (Olawade & Aienobe-Asekharen, 2025). AI adoption in Malaysia should therefore move beyond theoretical potential towards operational integration within customs enforcement systems.
Dunsin et al. (2024) highlight the growing need for practical AI-powered tools to combat increasingly sophisticated illicit activities. Nevertheless, the study raises serious concerns regarding bias, fairness, transparency and ethical implications. There is a growing need for frameworks to guide the development and deployment of AI applications in criminal justice, ensuring transparency, accountability, non-discrimination and respect for civil liberties (Situmeang et al., 2024). While AI poses potential risks, it also offers significant benefits; however, its application in criminal justice carries high stakes. Previous studies have examined AI applications, their characteristics and their implications in criminal justice systems. National law enforcement authorities acknowledge the potential of AI in investigating cybercrimes; however, a lack of specialised expertise and understanding of legal and ethical implications may hinder effective implementation (Custers, 2022; Llinares, 2020; Situmeang et al., 2024; Velasco, 2022). These concerns are important when considering AI implementation in illicit tobacco enforcement systems.
This study examines the potential applications of AI in addressing the ITT in Malaysia, highlighting its benefits, challenges and practical recommendations for implementation. This conceptual paper proposes an AI-driven enforcement framework tailored to Malaysia’s regulatory and enforcement context. This study contributes to three areas. First, it connects the theoretical realm of AI applications to the more applied world of customs enforcement systems. Second, it offers a Malaysia-centric enforcement lens by situating global AI tools within the context of localised regulatory and operational challenges. Third, it bridges the use of ML technologies with enforceable strategies to improve risk detection, monitoring and compliance in customs and tobacco enforcement systems.
2. Approach and methods
This study adopts a conceptual approach that integrates strategic policy analysis with a literature review. The proposed framework is informed by the synthesis of diverse materials, including peer-reviewed academic studies, enforcement data from government reports, international agency guidelines and documented case studies on anti-ITT interventions globally and regionally. This approach enables the identification of both systemic gaps and actionable opportunities within Malaysia’s current enforcement landscape. Rather than relying on primary empirical data, the methodology focuses on aligning AI capabilities with practical customs and regulatory challenges. The ultimate goal is to propose a practical, AI-driven enforcement framework suited to the Malaysian context for more effective and transparent governance in tackling the ITT.
3. Background and context
3.1. The scope of the illicit tobacco trade in Malaysia
In Malaysia, the illicit cigarette problem is proving to be a significant challenge affecting economic stability and law enforcement systems. Malaysia’s tobacco excise tax, aimed at reducing smoking prevalence by raising the price of cigarettes, has unwittingly given rise to a profitable black market. Often priced well beneath their legally taxed counterparts, smuggled cigarettes also have a special appeal for cost-sensitive consumers, particularly among low-income households (Kemp & Galemba, 2020; Lampe, 2010; Paraje et al., 2022). In the Malaysian enforcement context, these dynamics are intensified due to cross-border trade routes and price differentials with neighbouring countries.
The Illicit Cigarettes Study reported a substantial volume of illicit cigarette consumption in Malaysia, illustrating the scale of the illicit market (Nielsen Consumer LLC, 2024). Although these black market cigarettes are sold at lower prices than legally taxed products, enforcement seizures represent only a small proportion of the total illicit trade (Table 1). In Malaysia alone, it is estimated that illegal products control over 50 per cent of the tobacco market, highlighting the scale of the problem (Lim et al., 2018; Oxford Economics, 2019; Tan et al., 2019; Tay, 2019). Illicit tobacco detection is increasingly complex with methods used including falsifying shipping declarations, using modified transport vehicles with hidden compartments and taking advantage of porous border zones. These smuggling practices continue to challenge conventional enforcement and border monitoring systems (Kemp & Galemba, 2020). These increasingly sophisticated smuggling techniques complicate shipment screening, border monitoring and customs risk assessment.
3.2. Limitation of current enforcement efforts
The fight against the ITT in Malaysia is challenging and has significant cost implications. So far, most approaches have been manual, relying on resource-intensive inspections, random border checks and fixed monitoring systems. These measures have limited success and are constrained by human resources and legacy technology constraints. A report from the United Nations Environment Programme (UNEP, 2016) emphasises these limitations in conventional enforcement approaches. These limitations reduce detection effectiveness and highlight the need for more intelligence-driven enforcement approaches supported by AI-assisted surveillance and automated risk detection systems.
Another problem is the shifting and flexible structure of smuggling. Criminal organisations constantly adapt their methods to circumvent enforcement actions, using sophisticated logistics and technology to avoid detection. There is limited collaboration between the various enforcement agencies involved, and real-time data sharing is rare, further impeding the ability of these agencies to get ahead of the networks operating in this unregulated space. Conventional detection rarely penetrates the highly organised structure of illicit trade (Leyro, 2015). More adaptive and intelligence-led enforcement systems are needed to improve shipment monitoring and customs risk detection (Kupatadze, 2025).
4. The potential use of artificial intelligence
Marinova (2024) shows that AI enhances enforcement efficiency and detection capabilities. It provides a practical approach to improving enforcement tools and overcoming the challenges of current approaches. Using AI-powered tools enables enforcement agencies to shift from reactive to proactive, significantly improving their ability to crack down on the ITT (Kupatadze, 2025). In Malaysia, these approaches may strengthen customs surveillance and risk-based inspections. From an enforcement perspective, AI represents a transition from reactive inspection-based systems to proactive intelligence-led customs operations (Marinova, 2024).
Studies have demonstrated the use of AI-based tools in surveillance and anomaly detection (Broekhuizen et al., 2023; Montasari, 2022). ML algorithms are used to analyse large datasets, including customs records, surveillance footage and financial transactions, to identify anomalies that suggest smuggling activities. These systems enable customs authorities to prioritise high-risk consignments for targeted inspection. Predictive analytics is used to identify high-risk areas and determine when resources should be deployed. Apart from static surveillance methods, AI and real-time monitoring systems alert security officials immediately if any dubious activities are detected, thus minimising the chances of smuggling businesses going undetected. AI-based image recognition and NLP tools are used for document verification and vehicle inspections, streamlining the process and reducing human error (Broekhuizen et al., 2023; Montasari, 2022). These systems support automated cargo screening, anomaly detection and intelligence-led customs inspections.
The Royal Malaysian Customs Department (JKDM) installed 40 AI-powered baggage scanners worth MYR40.72 million at key entry points nationwide to enhance security (Bernama, 2024). These scanners improve image analysis and contraband detection, particularly for narcotics. Since their implementation, JKDM detected multiple smuggling attempts, including MYR577,625 worth of cannabis-related drugs and MYR492,375 worth of heroin (Table 2). Additionally, MYR65,160 in undeclared cash was intercepted. The procurement includes maintenance through leasing arrangements, with 66 scanners now operational across Malaysia (Bernama, 2024). The Malaysian customs scanner initiative illustrates the operational role of AI-assisted cargo inspection systems.
AI-driven systems are also used globally in enforcement contexts for customs and border security operations, including cargo risk profiling, automated inspection and fraud detection, improving detection efficiency and reducing reliance on manual processes (Marinova, 2024). AI customs enforcement is also applied at regional customs operations (Cao & Zheng, 2024). The World Customs Organization’s (WCO) Cargo Targeting System (CTS), for example, provides the means for risk-based cargo profiling and intelligence-led inspections (WCO, n.d.). The European Union has a Customs Risk Management System (CRMS) that allows member states to share information in real time, leading to coordinated risk assessments (European Commission, 2021). China’s Smart Customs initiative leverages big data analytics and artificial intelligence for predictive risk detection and automatic clearance procedures (Cao & Zheng, 2024). These systems provide empirical evidence of AI deployment in operational customs environments. AI is increasingly integrated into cargo screening, border monitoring and customs risk management operations.
Existing literature highlights key implementation challenges (Liber et al., 2015; Nilgiriwala et al., 2024). Integrating AI into contemporary enforcement frameworks may strengthen the abilities of authorities to respond to smuggling strategies. However, additional investment in infrastructure, training and cross-agency collaboration is required for the effective implementation of AI. Addressing these implementation barriers is essential for improving enforcement effectiveness and customs intelligence capabilities in Malaysia (Liber et al., 2015; Zainol et al., 2019). While AI has improved detection efficiency and operational performance in enforcement contexts, challenges such as false positives, high implementation costs and system integration issues remain. The proposed framework integrates predictive analytics, customs intelligence, cargo risk profiling and regional enforcement coordination.
Despite its operational advantages, AI-driven enforcement systems remain vulnerable to false positives, algorithmic bias and adaptive smuggling strategies. Criminal networks may alter shipment routes, concealment methods or transaction patterns to evade predictive detection systems. These limitations highlight the need for continuous model retraining, human oversight and balanced integration between AI-assisted systems and conventional customs intelligence operations.
4.1. Opportunities for using artificial intelligence
Kupatadze (2025) and Marinova (2024) report the role of AI in enhancing enforcement efficiency. Implementing AI within enforcement strategies has brought unique solutions to these challenges. AI processes large amounts of data, identifies patterns and provides real-time predictions improving enforcement efficiency and supporting targeted interventions against smuggling networks (Kupatadze, 2025; Marinova, 2024; Olawade & Aienobe-Asekharen, 2025). This section synthesises these findings and outlines key opportunities for AI application in ITT enforcement within customs and border monitoring systems.
4.1.1. Predictive analytics for risk analysis
Marinova (2024) and Kupatadze (2025) highlight the use of AI-driven predictive analytics in enforcement contexts. AI-powered predictive analytics and risk assessment models are increasingly used in customs enforcement operations. These systems analyse vast amounts of data, including shipment records, surveillance footage and intelligence reports, to identify high-risk individuals, transportation routes and vulnerabilities in the supply chain (Kupatadze, 2025). ML-based risk profiling systems may also assist customs authorities in prioritising high-risk consignments for targeted inspection. By being vigilant, law enforcement is better equipped to use resources strategically to anticipate potential smuggling activities and intercede before goods can be successfully smuggled. Malaysian customs authorities may use these systems to prioritise high-risk consignments and border inspections.
4.1.2. Real-time border surveillance
By utilising satellite imagery, sensor networks and other AI-driven technologies, organisations enhance their ability to detect and track the movements of suspicious cargo and vehicles in real time. This improves law enforcement’s capacity for intercepting illegal tobacco shipments and disrupts smuggling networks. AI analyses data from multiple sources including satellite imagery and surveillance records to identify abnormal movement patterns and provide actionable intelligence (Cao & Zheng, 2024). Incorporating these technologies into enforcement strategies can provide law enforcement with improved situational awareness and the tools necessary to effectively disrupt the cross-border movement of illicit tobacco products.
Evidence also supports the use of anomaly detection in identifying illicit trade patterns (Kupatadze, 2025). AI-based intelligent anomaly detection algorithms have the potential to mine large volumes of trade, financial and logistics data to identify patterns of noncompliance concerning illicit tobacco activities (Kupatadze, 2025). By analysing large datasets using cutting-edge ML methods, these systems can identify unusual transactions, transportation routes or anything else that requires further investigation by law enforcement agencies. Implementing AI-driven anomaly detection can considerably boost the effectiveness and efficiency of efforts to combat the illicit market for tobacco by automating the detection of these unlawful activities (Marinova, 2024). AI-based detection systems may improve shipment targeting and anomaly identification across high-risk trade routes.
4.1.3. Monitoring and disruption of online sales of contraband tobacco
AI-powered web crawlers and NLP techniques are used in constantly monitoring the selling of illegal tobacco products on the internet (Chaloupka et al., 2015; Olawade & Aienobe-Asekharen, 2025). These systems scan e-commerce platforms, social media channels and other online marketplaces to identify suspicious listings, keywords and user behaviours that may indicate participation in the ITT (Mackey et al., 2017). By continuously monitoring these digital channels, authorities can quickly target and shut down the e-commerce operations of smugglers and other bad actors seeking to distribute contraband tobacco products (Li et al., 2019).
AI-enabled NLP is also used to detect suspicious textual content and contextual patterns in online listings and activities. Such systems can analyse larger volumes of data in a structured manner through ML algorithms and match them with trade activities to detect suspicious or irregular patterns, even highly sophisticated activities (Mackey et al., 2018). Authorities may use these outputs to disrupt online tobacco distribution networks (Küçükali & Erdoğan, 2025).
In addition, these tools use computer vision techniques to identify suspicious images and visual cues of illicit tobacco products. Based on product images, packaging and all other visual aspects, the system can improve detection efficiency and identify suspicious online distribution channels. The model incorporates NLP and computer vision as a part of a multimodal approach for detecting and disrupting the illegal online sale of contraband tobacco. AI systems are designed to identify patterns through machine learning, with human feedback from law enforcement and other stakeholders allowing the machine to improve its performance over time (Bachhuber & Merchant, 2017). This iterative process allows the system to adapt and stay relevant to the changing strategies and methodologies used by smugglers and online vendors of illegal tobacco goods (Chaloupka et al., 2015).
4.1.4. Unlocking citizen power in the fight against illicit tobacco trade: A public reporting platform
AI-enabled public reporting platforms can strengthen collaboration between citizens, enforcement agencies and the private sector in combating the illegal tobacco market. In the proposed framework, citizens could submit information through a mobile application or website, with AI-assisted tools used to process and organise those reports. Such a system could identify patterns and trends in user-submitted reports to support targeted enforcement actions. Analysis of citizen-reported data may improve real-time identification of emerging illicit tobacco hotspots and suspicious distribution activities (Chaloupka et al., 2015). Additionally, the public reporting platform functions as a two-way communication channel where authorities provide feedback and updates to citizens on the actions taken in response to their submissions.
By doing so, this public reporting system reinforces the partnership between the public and private sectors by directing citizen-generated intelligence towards law enforcement investigations. The findings can guide and rationalise the distribution of resources to achieve a more complete and consequential interception of the ITT (Broekhuizen et al., 2023). However, challenges such as ensuring data privacy, preventing platform misuse and integrating reports into enforcement frameworks remain crucial considerations (Bachhuber & Merchant, 2017). Data privacy must be safeguarded to protect individuals submitting reports, while security measures are necessary to prevent false or manipulated reports from undermining enforcement efforts. To address these concerns, approaches such as federated learning, edge computing, zero data retention and secure in-country data systems can be implemented to ensure privacy protection and improve trust in AI-assisted public reporting systems.
5. Challenges in implementing artificial intelligence
The potential of AI as a tool to combat the ITT in Malaysia is significant; however, to ensure that these transformative technologies are implemented effectively, responsibly and ethically in enforcement efforts, the key challenges associated with their application must be addressed. Addressing these barriers is essential to harnessing the power of AI to disrupt smuggling networks and improve the effectiveness of enforcement operations.
5.1. High initial costs
The high cost of implementing emerging AI technologies is a significant barrier, especially for public law enforcement agencies operating under tight budgetary constraints. Building the required infrastructure means obtaining advanced high-capacity hardware like high-performance servers and cutting-edge tracking technology. Agencies must also purchase advanced software platforms to deliver complex AI algorithms and process massive data sets. Moreover, substantial investment in staff training is required to ensure that enforcement staff can operate and maintain these cutting-edge systems (Administrative Conference of the United States, 2024).
The initial costs of adopting AI, from hiring skilled personnel to purchasing powerful servers, can limit the extent of adoption. Considering this challenge, phased and strategic deployment approaches may help reduce implementation burdens. Focusing on the most high-risk areas and introducing AI tools in a phased manner can help balance costs while improving enforcement effectiveness in disrupting smuggling networks (Sharma et al., 2020). Continued funding and strategic public-private partnerships (PPPs) may be needed to effectively implement AI to tackle the ITT. Sufficient funding and collaborations with private sector technology companies can help enforcement agencies overcome financial and technical barriers associated with AI-enabled enforcement systems (Administrative Conference of the United States, 2024; Velasco, 2022).
5.2. Data privacy concerns
AI relies on large-scale data collection and monitoring; therefore, it raises legitimate ethical questions regarding personal privacy and individual liberties (Olawade & Aienobe-Asekharen, 2025). For example, real-time monitoring of people, vehicles, and even online activity can potentially violate fundamental rights and freedoms if adequate caution is not exercised and individual privacy is not prioritised in system design. This problem is especially prevalent in enterprise systems that interpret high-stakes personal data, including payment histories, communication logs or other confidential information (Ejeofobiri et al., 2024). Clear regulatory guidelines are needed to define the scope and limits of AI-related data usage. Transparency and accountability measures, alongside strong, layered data protection safeguards, will help build public trust and confidence in the responsible adoption and implementation of AI systems (Araujo et al., 2020; Gillespie et al., 2023). Moreover, establishing robust and independent oversight mechanisms becomes paramount to ensure that AI tools are deployed ethically and in the public interest (Petković, 2023). Practical solutions include federated learning, where data remains within local systems while only model updates are shared; edge computing to process data locally; zero data retention policies; and deployment within secure in-country data centres to ensure privacy protection and regulatory compliance.
5.3. Data integration with the existing systems
Legacy systems often do not integrate well with modern AI technologies (Marinova, 2024). Integrating AI tools into these existing frameworks is technically challenging and resource intensive. Different databases that are not linked, fragmented channels of communication, and inconsistent technologies increase implementation complexity (Berk, 2020; Broekhuizen et al., 2023). Major system upgrades are needed to overcome these significant barriers. Agencies must modernise their IT infrastructure to include robust data management systems and interoperability across enforcement systems. It will take significant planning, resource allocation and coordination to successfully transition to modern systems (Saenz et al., 2023; Zuiderwijk et al., 2021). Additional training for enforcement personnel is also needed to address skill gaps to operate and maintain these advanced AI-powered tools effectively. Developing the required technical capabilities and nurturing a culture of innovation in the workforce is critical in realising the full potential of AI in improving enforcement efficiency (Araujo et al., 2020; Berk, 2020; Broekhuizen et al., 2023). Working with established technology vendors and domain experts can also support a smoother integration process. External partners can highlight best practices in this complex landscape of developing AI solutions while aiding with technical support. Such expertise can significantly reduce implementation challenges and enable smoother implementation of AI-based tools within the current enforcement system (Velasco, 2022; Zuiderwijk et al., 2021).
5.4. Risk of misuse
Additionally, the use of AI carries significant risks of misuse, whether through intentional abuse or improper applications. Without appropriate oversight and strong safeguards, AI tools could be misused for personal, political or commercial benefit, leading to outcomes such as false accusations, selective enforcement or corruption (Butcher & Beridze, 2019). Biased algorithms or poorly monitored surveillance systems, for instance, could lead to over-policing and inequitable profiling of specific demographics or individuals and undermine trust in enforcement efforts (Broekhuizen et al., 2023; Gillespie et al., 2023).
Thus, to mitigate these risks, comprehensive ethical frameworks and robust accountability mechanisms must be implemented to govern the deployment and use of AI technologies. These measures should include independent audits, transparent decision-making processes, and clearly defined responsibilities and oversight mechanisms for AI systems (Asaro, 2016; Broekhuizen et al., 2023). These AI tools should undergo regular and continuous monitoring and evaluation to assess their performance and identify potential vulnerabilities or biases to ensure alignment with the legitimate aims of law enforcement and ethical standards. These safeguards can support responsible and transparent AI governance while protecting civil liberties and the public interest in ITT enforcement (Gillespie et al., 2023).
6. Call to action
To fight the national and international ITT effectively and maximise AI’s potential, Malaysia must implement a strategic, collaborative and future-ready response based on a holistic and integrated approach. This requires the integration of emerging technologies, strong governance frameworks, ethical guidelines and the engagement of the wider community. AI-driven solutions in enforcement strategies support efforts to counter the ITT. Utilising AI to process enormous datasets, recognise patterns and provide real-time insights, enforcement agencies can improve their effectiveness, dismantling smuggling networks more efficiently. Yet the effective deployment of these technologies depends on a systems approach addressing the preconditions for their successful and responsible use.
6.1. Higher level of government investment
Substantial government investment is critical to promote the widespread adoption and successful implementation of AI-powered solutions in enforcement against the ITT (Oxford Economics, 2019). The money should be used to purchase advanced surveillance equipment, including AI-driven drones, thermal imaging cameras and automated monitoring platforms, which can enhance enforcement agencies’ detection and tracking powers (Table 2; Aslett, 2024; Cao & Zheng, 2024). Another critical aspect is upgrading the IT infrastructure to ensure that these advanced technologies work well with existing systems and can provide real-time analysis of large data sets to identify patterns and deliver timely insights. In addition, specialised training programs will be required to develop the capacity of enforcement personnel who will operate and manage these systems to utilise them effectively. Up-skilling local talent and allocating resources to targeted R&D that address Malaysia’s unique smuggling challenges will also be crucial in realising the transformative potential of AI applications to combat the ITT (Aslett, 2024; Oxford Economics, 2019).
6.2. Public-private partnerships
Public enforcement agencies working with private stakeholders are essential for the effective deployment of AI in the fight against the ITT. Technology providers can offer advanced technologies, like ML platforms or AI solutions to enhance surveillance (Velasco, 2022). These technologies can improve coordination, shipment tracing and enforcement responses to smuggling activities. Tobacco manufacturers and logistics companies also have an essential role in integrating AI-driven tracking systems within their operations. They can enhance accountability and minimise opportunities for illicit diversions, as the real-time monitoring and traceability provided by these technologies can also facilitate the detection and disruption of smuggling (Paraje et al., 2022). Intelligence sharing is essential to enhance PPPs so enforcement agencies have better visibility into smuggling networks and can take targeted action to disrupt them. Establishing such joint actions will help maximise the impact of AI-powered solutions in the fight against the ITT (Marinova, 2024).
6.3. Public awareness campaigns
Public awareness campaigns are important to solicit active community engagement and support in fighting the ITT. By educating citizens on the economic and social impacts associated with this crime, such campaigns can help create a shared, collective sense of responsibility and commitment to address this issue (Joossens et al., 2010; Paraje et al., 2022). A key means by which even ordinary public members can contribute will be a proactive endorsement of reporting anything suspicious through accessible AI-enabled mobile apps and anonymous hotlines (Chaloupka et al., 2015; Küçükali & Erdoğan, 2025). These AI-enabled reporting systems can support enforcement agencies by improving real-time information sharing and community-based monitoring. This approach strengthens collaboration between the public and enforcement authorities in combating the ITT (Joossens et al., 2010; Paraje et al., 2022).
6.4. Regional collaboration
Since the ITT is highly complex and transnational, effective collaboration is crucial to address this issue on a regional level. In this context, Malaysia needs to collaborate with its neighbours to implement strong intelligence-sharing mechanisms proactively, align AI-enabled enforcement tools, and join forces to devise holistic approaches to curb cross-border smuggling of tobacco products (OECD, 2016). Dedicated regional task forces and training can ensure that regional enforcement agencies leverage their collective knowledge and ground strength to disrupt the operational enterprise of smugglers more effectively. A coordinated regional response grounded in trust and intelligence sharing can strengthen enforcement effectiveness against illicit trade networks. Such a collective and trans-border initiative can, in its essence, improve the effectiveness of enforcement actions (Joossens et al., 2010).
6.5. Frames of reference for ethics and regulations
The analysis and application of AI models must align with ethical and regulatory frameworks to ensure responsible, transparent and accountable development and deployment of AI-powered solutions in the fight against the ITT (Aslett, 2024). Comprehensive policies should address data privacy, the appropriate scope of AI applications and mechanisms for accountability to reduce potential misuse (Bachhuber & Merchant, 2017). Independent audits will be necessary to ensure public trust in AI systems and ensure rigorous adherence to ethical principles and privacy protections. Establishing clear, unambiguous principles on the governance of AI-enabled enforcement tools is urgently needed to ensure equitable use while respecting rights. The framework should describe the roles and responsibilities of stakeholders and empower citizens with the tools and venues to raise their concerns and keep authorities accountable.
Through this holistic, multi-tiered strategy, Malaysia can harness the power of AI to address the underlying networks that sustain illegal tobacco trafficking, ensure the health and safety of its individuals, and reclaim vast amounts of missing tax revenue, all of which contribute to the preservation of the country’s economic stability and national security (Chaloupka et al., 2015). However, these efforts must maintain public trust, accountability, and responsible AI governance while protecting individual rights (Aslett, 2024; Küçükali & Erdoğan, 2025).
7. Conclusion
The ITT continues to undermine Malaysia’s economic stability and border security. While AI offers significant potential to improve enforcement through predictive analytics, real-time surveillance and data-driven decision-making, its success depends on coordinated policy support. Government investment is essential to build the infrastructure and technical capacity needed for AI integration. At the same time, PPPs can accelerate innovation and improve operational outcomes. AI-enabled public reporting systems and community engagement initiatives can support enforcement monitoring, while regional cooperation is crucial for addressing cross-border traffics and intelligence sharing. Finally, ethical frameworks must be established to ensure responsible AI use, protect privacy and maintain public trust. By aligning these efforts, Malaysia can advance towards a more transparent, secure and technology-driven approach to tackling the ITT challenge. Without such integration, enforcement efforts risk remaining reactive and insufficient against increasingly sophisticated illicit trade networks. Malaysia’s future customs enforcement capacity will increasingly depend on the integration of predictive AI systems with regional intelligence cooperation and risk-based cargo management.