AI-Assisted Detection of Digital Hundi and Online Gambling-Linked Money Laundering: Evaluating the Legal Sufficiency of Bangladesh's New Gambling Prevention Act 2026 and MLPA 2012

NAZIA RAHMAN; MOHAMMAD MINHAZUR RAHMAN

Open Access Research Article White Black Legal – International Law Journal · ISSN 2581-8503

AI-Assisted Detection of Digital Hundi and Online Gambling-Linked Money Laundering: Evaluating the Legal Sufficiency of Bangladesh's New Gambling Prevention Act 2026 and MLPA 2012

Author(s): NAZIA RAHMAN MOHAMMAD MINHAZUR RAHMAN

Volume Volume 4 Issue VOLUME 4 ISSUE 2 Published October 2026 Pages 1278-1310

Download PDF Google Scholar

Read the Full Research Paper

Access the complete open-access article in PDF format. No login is required.

Open PDF

Abstract

AI-ASSISTED DETECTION OF DIGITAL HUNDI AND ONLINE GAMBLING-LINKED MONEY LAUNDERING: EVALUATING THE LEGAL SUFFICIENCY OF BANGLADESH'S NEW GAMBLING PREVENTION ACT 2026 AND MLPA 2012

AUTHORED BY - NAZIA RAHMAN

Affiliation/Current Position: State University of Bangladesh

CO-AUTHOR - MOHAMMAD MINHAZUR RAHMAN

Affiliation/Current Position: Lecturer, Department of Law, State University of Bangladesh

Abstract

Research Background: The fast-growing mobile financial services, online betting, and cross-border digital payment systems in Bangladesh facilitate the distribution of winnings from online gambling in hundi groups, hidden in digital currency. The Gambling Prevention Act 2026 addresses this by introducing criminal laws against online gambling and online betting, along with the facilitation of gambling and betting via the transfer of funds, using cryptocurrencies, and with the help of artificial intelligence.In response, the Gambling Prevention Act 2026 criminalises the use of AI, cryptocurrencies and their facilitation, as well as the transfer of funds for gambling purposes, online gambling and online betting.

Purpose of the Research: This article examines the extent to which the 2026 Act, when read with the Money Laundering Prevention Act 2012 (MLPA), does provide a lawfully adequate mechanism for the detection, investigation and prosecution of digital-hundi and gambling-related money laundering.

Data, Materials and Methodology: A qualitative doctrinal and policy analysis is performed on the two statutes and constitutional safeguards, Bangladesh Bank and Bangladesh Financial Intelligence Unit documents, FATF standards, reported typologies of enforcement, and peer-reviewed research on machine-learning-based anti-money-laundering systems. A statute-to-standard matrix assesses substantive coverage, institutional powers, evidential reliability, procedural safeguards, data governance and international cooperation.

Results: The 2026 Act, the previous loophole in the offence definition is significantly filled with the introduction of the gambling modalities and the incorporation of the MLPA predicate-offence framework, account and asset restraint, and the use of AI, risk scoring, transaction monitoring and data analytics. However, the framework does not explicitly require validation of the model, explainability, prioritising data minimisation, restrictions on the data retained, independent audit, human review, a right to be corrected and specific remedies if a person is incorrectly blacklisted, blocked or frozen.

Major Findings: Bangladesh has a robust substantive enforcement gateway, and a partial legal sufficiency. The effectiveness and constitutional longevity of these AI outputs will rely on the ability to incorporate financial intelligence from MLPA and reviewable evidence and clear protections against over-surveillance and false positive enforcement by judicial actors.

Keywords: artificial intelligence; digital hundi; online gambling; money laundering; Bangladesh

JEL Codes: K14; K23; K24; G28

Chapter 1: Introduction and Basic Hypotheses

1.1 Background and Regulatory Transition

Today, the issue of Gambling Control in Bangladesh is in the cross-roads of constitutional policy, financial regulation, cyber enforcement and data governance. Article 18(2) of the Constitution is a policy direction in favour of abstention from gambling that sits within the broader constitutional obligations of protection of law, personal liberty, expression and privacy. The main statute for the past 100 years has been the Public Gambling Act 1867, which was a colonial enactment in the context of common gaming houses and physical instruments. It was unstructured and was not suitable for all the components of remote casinos, mirror websites, app casinos, cryptocurrency wallets, mobile financial services (MFS), identity fraud and transnational payment chains.^{^[1]}

The Gambling Prevention Act 2026 repeals the 1867 Act and introduces a tech-based approach to criminality. Its definitions are applicable to online or remote gambling, various online betting types, digital gambling platforms, digital wallets, cryptocurrency or digital assets, fake SIMs, fake MFS accounts, biometric forgery, mirror gambling sites or match fixing and facilitation. The two sections of the act that prohibit online gambling and online betting provide distinct prohibitions, the three sections of the act that prohibit transference, concealment, legalisation and conversion of gambling funds through banks, MFS, digital wallet, hundi, or other financial systems are distinct, and the predicate-offence structure of the MLPA is applied to these three sections. This is a major change in the suppression of a location to the tracing of a network.^{^[2]}

1.2 The Digital Hundi-Gambling Nexus

Digital hundi is no longer a one-off payment product. It is an informal value transfer mechanism which can include domestic collection via MFS, settlement via overseas agents, trade or remittance offsets, cryptocurrency conversion or payments to foreign platform operators. The on-line gambling industry is ideally suited to this architecture since small repeated values can be utilized for deposits, transferred in via a merchant or individual account, deposited in settlement accounts and then forwarded to actors with virtually no connection to the business function. The same chain can be reversed as winnings or agent commissions are given out.^{^[3]}

The instance comes from the Bangladesh Financial Intelligence Unit's case study of 2023-24. 51 MFS accounts belonging to 50 "legitimate businesses" were connected to a gambling application named 'Z Casino' in an MFS provider's suspicious transaction report.An MFS provider's SUSI matched 51 MFS accounts, belonging to 50 apparently legitimate businesses, with a gambling application called 'Z Casino'. About BDT 1,122.5 million was collected in one year, usually in small amounts like BDT 100, BDT 200, BDT 500 and BDT 600 are introduced in a repetitive manner. Ten trust-cum-settlement bank accounts were set up and transferred to the profiles of the recipient who had inconsistent activities including remittance beneficiaries, cattle farms, gold businesses, and traveling agencies. BFIU was apprehensive about the possibility of onward settlement through the channel of ‘hundi' or ‘hawala' and disseminated intelligence about 50 persons/ entities. Transaction by transaction rules is not sufficient because the laundering signal occurs in velocity, repetition, account relationships, business inconsistency and cross channel movement.^{^[4]}

In addition, there is increasing reporting burden. During FY2023-24, BFIU received 17,345 suspicious transaction and activity reports, while the figures of online gambling, betting, foreign exchange and cryptocurrency trading, and digital hundi were amongst the factors driving up the number of reports in FY2024-25, which stood at 30,199. That's not a rule of thumb that more volume equals more effectiveness. It could also build up huge alert queues, or work inconsistently from alert to alert, escalating to pressure on to use ambiguous scores. The central legal issue is not just that the State should be able to use AI, but how automated detection becomes justifiable suspicion, reasonable action, and admissible evidence.^{^[5]}

1.3 Problem Statement

Section 44 of the 2026 Act mandates the creation of AI-driven monitoring systems, deep packet inspection, risk scoring, transaction monitoring, data analysis, and more, all designed to combat online gambling and betting activities. Bangladesh Bank had already directed payment-system operators to implement AI-based technology to continuously monitor merchants and customers they suspected to be involved in online gambling and to take necessary action if involvement is found. The powers provide a legal framework to enable adaptive detection. They also do not set standards for training data, tolerable false-positive rates, explainability criteria, validation frequency, acceptable data sources, retention periods, or thresholds or processes for allowing the challenged individual to contest an inaccurate score, blacklist listing, account restriction.^{^[6]}

Enforcement of the Act is an added concern. It allows for emergency warrantless search and seizure when evidence is likely to be transmitted or destroyed, Website/Platform blocking, closure of financial accounts and establishment of a national digital blacklist of identity, account, device, domain and network data. They might be required to combat the swift movement of gambling infrastructure, however they can impact permitted players, merchants and communications. Article 43 guarantees the privacy of home, correspondence, and communication, with the proviso of reasonable legal restrictions and Bangladesh's reasonableness jurisprudence demands a rational relationship between a restriction and the threat invoked. Operational capacity must therefore be accompanied by legal safeguards that can guarantee that it is not used in an arbitrary or over-broad manner.^{^[7]}

1.4 Research Questions and Hypotheses

· Does the 2026 Act adequately define substantive crimes and investigative tools to be able to catch the digital hundi and laundering of money through online-gambling?

· Does its engagement with the MLPA provide a valuable pathway from AI generated indicators to financial intelligence, prosecution and confiscation?

· Are the existing safeguards of surveillance, data and procedure sufficient so that the enforcement of AI is constitutionally proportionate and institutionally accountable?

Three propositions are tested in the analysis. H1 is that the 2026 Act substantially fills in the holes in substantive offence and predicate offence legislation that made it difficult to enforce remote gambling networks. H2 is that the legal sufficiency is not complete, because the powers to block, blacklist, freeze and search are not governed by an algorithm and do not provide standards for a remedy. H3 is about a need to cooperate between BFIU, Bangladesh bank, payment service providers, telecoms regulators, law-enforcement agencies, prosecutors and courts, and that interoperable, risk-based cooperation will be more important than any one AI model.^{^[8]}

1.5 Contribution and Structure

Researchers typically view online-gambling laundering, informal value transfer and AML machine learning and digital-rights protections as distinct issues. This article reviews and incorporates them into a Bangladesh specific legal sufficiency test. It considers the substantive sufficiency (whether the conduct is criminalised), institutional sufficiency (whether competent authorities can obtain information, and this can be shared), technological sufficiency (whether the relevant patterns can be detected), procedural sufficiency (whether coercive action is reviewable) and rights-based sufficiency (whether data use is necessary, proportionate and contestable). This framework is developed in the following literature review; methodology is explaining the doctrinal matrix; Results are evaluating the Acts and a representative BFIU typology; and the final part is policy implications, limitations and conclusions.^{^[9]}

Chapter 2 Literature Review and Conceptual Framework

2.1 Online Gambling as a Laundering Environment

Online gambling platforms can carry out the typical laundering activities of placement, layering and integration without being present in a physical casino. A customer can fund one or multiple accounts, go through intermediaries to deposit value, place offset bets, withdraw balances as seemingly legitimate winnings, and trade platform credits or work through affiliates and agents to complete trades outside of the platform. The marginal cost of creating accounts is low and operators are available across the globe, which makes the number of layers possible. Fiedler states that online gambling can reduce the cost and scale of laundering due to the separation of identity, location, ownership and payment channels. The following shouldn't be taken as a rule: Regulated operators that perform effective customer due diligence checks, source-of-funds checks, and transaction monitoring can still generate useful intelligence. In a series of interviews with compliance and integrity staff from the regulated side of the British sector, Brooks found that a gambling platform could be an infrastructure used to launder money or a platform that would be capable of acting as a gatekeeper for the anti-money laundering process.^{^[10]}

So the principle weaknesses are not intrinsic but relational. These include limited real play, rapid deposits and withdrawals, various funding options, third-party payments, frequent device or location changes, collusive or low-risk wagers, transfers between linked accounts, bonus and affiliate programs, and fiat currency to e-wallet balance and virtual asset conversions. The added dimension is electronic smurfing: a big amount can be made up of numerous small transactions, without attracting thresholds or human notice. Thommandru sees online betting and electronic payments and cryptocurrency as mutually reinforcing channels to move broken value across institutions and borders. The patterns are reminiscent of digital hundi collection, in which a large proportion of local payments are matched against an obligation in an offshore currency.^{^[11]}

2.2 Digital Hundi, MFS and Networked Concealment

Conventional explanations of hundi have focused on the trust that is involved in the settlement of the brokers. The evidential surface is modified by digitalisation, but not the economic function. Local collectors are able to take advantage of personal, agent or merchant MFS accounts; settlement is through bank transfer, import or remittance offsets, cash, virtual assets, or accounts of nominees. The network is resilient as no single transaction must contain a description of the gambling or cross the border. The duty that crosses the border is between brokers. This distinction is important doctrinally because it is necessary to establish a connection between the domestic payment activity and the predicate conduct, knowledge and intent, and the MLPA's conceptions of "proceeds," "movement," "concealment," "possession," and "use or participation.^{^[12]}

The BFIU 'Z Casino' typology provides a model of the concrete. Businesses that formally described what they did as legal collected small amounts of payment; they put money into settlement accounts and drew out money into the hands of recipients that were not consistent with the formal description of their activities; suspected hundi dealers then fulfilled the function of the cross-over. The usual rule of thumb that would only pick up unusual large transfers would lose a lot of this chain. Better metrics include repeated round-value deposits, links with known agents, uncommon device identifiers, common beneficiaries, rapid pass-through, unusual cash-out geography, conversion to virtual assets, and merchant-category inconsistency. The unit of analysis should be the account/device/merchant/beneficiary graph, not an individual payment.^{^[13]}

2.3 The MLPA Architecture and the Predicate-Offence Principle

The MLPA outlaws the knowing movement, conversion or transfer of property to conceal the proceeds, or to help in the evasion of reporting; the movement across borders to conceal the illegality; the structuring to avoid reporting; the acquisition, possession or use of proceeds; the concealment; and participation, conspiracy, attempt or abetment. It outlines suspicious transactions as those that deviate from normal activities, as well as suspicious circumstances relating to the proceeds or terrorist financing. It also provides BFIU with powers to accept and assess reports, request information, make temporary suspensions/freeze and to monitor the reporting organisations and share intelligence. Reporting organisations are required to keep records of their customers' identities and transactions and to make suspicious transaction reports of their own accord. ^{^[14]}

This architecture calls for a predicate offence or other qualifying criminal source, but doesn't mandate that one first be convicted of the predicate offence to be convicted of money laundering. Therefore, the section 25 of the 2026 Act must be worded correctly. It eliminates ambiguity on whether the full MLPA can be invoked in online-gambling cases, by treating gambling-fund transactions and the use of money laundering and cryptocurrency as predicate offences. The formulation, however, poses interpretive issues. In the title to section 14, the concealment or legalisation of funds for the purpose of gambling appears to be part of the conduct that triggers the separate laundering charge in section 25 of the MLPA. The prosecution will have to show that the crime of the “gambling-fund offence” is different from the later or multiple “laundering” actions in order to avoid charging the same crime more than once.^{^[15]}

2.4 Risk-Based AML and Technological Detection

FATF's risk-based approach is intended to enable jurisdictions and reporting entities to recognise, understand and address the risks of money laundering that they face, not necessarily in the same way. The recommendations address criminalisation and confiscation, suspicious transaction reporting, the functions of the FIU, international cooperation, new technologies and virtual assets, customer due diligence and investigative powers. The model is capable of supporting AI; as machine learning can prioritize heterogeneous transactions by context. It is also a legal restriction – risk scoring shouldn't be a blanket assumption of guilt.^{^[16]}

Supervised models can learn from the past transactions that have been investigated or reported; Unsupervised models can find anomalies without reliable labels; Graph Analytics models can be used to analyse relationships among Accounts, Devices, Merchants and Beneficiaries; Narratives from adverse media or Investigator Notes can be structured using NLP. Jullum and colleagues also found that a supervised model trained with information on transactions and investigations being reported and on investigation outcomes could outperform an existing bank process in prioritisation, and that the selection of the training data (alerts that were reported as well as ordinary transactions) had a significant impact on the model performance. The client-risk profiling process is also separated from the suspicious-behaviour flagging process, and limited public data, interpretability and fairness continue to be important research issues, as stated by Jensen and Iosifidis. ^{^[17]}

A hybrid system is more defendable than a completely automated classifier for digital hundi and online gambling. Rules can include legal defined red flags like frequent small depositions, quick pass-through or connections to blocked domains. Indirect relationships can be discovered with graph models. New behaviour may be detected by Anomaly models and past outcomes can be used to prioritise alerts through supervised ranking. Human investigators are then given the opportunity to ascertain whether there is an innocent explanation to the pattern, whether more information is needed and whether the statutory threshold for reporting, freezing and/or investigation is met. FATF's technology guidance encourages innovation while emphasising data quality, privacy, explainability, governance and ongoing human skills. ^{^[18]}

2.5 Explainability, False Positives and Evidential Reliability

It is hard to measure the accuracy of AML because the actual type of laundering that goes undetected is not known. A model might be imprecise when compared to the reports, but it still simply repeats what was investigated last. A very common feature of labeling is that a suspicious transaction report is not a judicial determination of guilt, and an unreported one is not necessarily illegal. Concept drift also occurs because of criminal behaviour. Too high a class imbalance can lead to a lot of false positives, too low a threshold a lot of false negatives. When the model output leads to closure, blacklisting, search, arrest, and/or reputational harm, there are legal implications from these limitations.^{^[19]}

Explainable AI can give local explanations of a specific score and global insights of a model's overall behaviour. In addition to intrinsic explanations, Narasapuram and associates also present post-hoc explanations to enhance auditability and investigators' confidence in suspicious-transaction detection. But explanation does not provide a foolproof protection. An explanation that is technically correct can still use an illegally accessed data source, or a proxy that discriminates against the source, or a correlation that has nothing to do with the source. The authors of Vilone and Longo's review argue that choosing the right explainability method is almost as critical as choosing the right tool, as there are a range of methods with varying scope, fidelity and evaluation. An explanation for legal purposes needs to be able to describe the material transaction characteristics and network connections, the model version involved, the relevant thresholds, the data provenance, and the human reasoning that led to the connection of the signal to the elements of the statute.^{^[20]}

2.6 Privacy, Surveillance and the Rule of Law

The information that AI transaction monitoring typically collects is information that the regulated institutions already have, while deep packet inspection could be used to analyze network content or metadata at a telecommunications checkpoint. The latter can be more intrusive as it can show communications not associated with the gambling. The Constitution guarantees the privacy of correspondence and of other communication, and article 17 of the ICCPR ensures that individuals shall have the right not to have their privacy intruded upon in an arbitrary or unlawful way. There are some legitimate aims for which restrictions can be justified, but these must have a legal basis, a rational connection, be necessary and contain safeguards against abuse. Oali Ahad is always invoked for the rule that a restriction has to have a direct connection to the public-order threat on which it is based. ^{^[21]}

The broad use of the term 'technology' in the 2026 Act ensures that the legislation is not dependent on any single technology, but also leaves some important issues to implementation. There has already been commentary in the contemporary press about how section 44's DPI authority and section 36's urgent warrantless powers could be used for too much surveillance. Any sustainable system must differentiate between the four concepts of population-level scanning, content versus metadata, detection versus enforcement, and temporary preservation versus retention. It should also mandate access control, logging, independent review and remedies.^{^[22]}

2.7 Conceptual Framework: A Five-Dimensional Sufficiency Test

The article is a synthesis of risk based regulation and a rule of law approach to automated enforcement. The five dimensions are used to determine the legal sufficiency of the dependent concept. The first question of substantive sufficiency is whether there is adequate clarity in the definition of gambling, facilitating, fund-transfer, virtual-asset, corporate and laundering conduct. Secondly, the question of institutional sufficiency is whether there is a coordinated power and duty of BFIU, Bangladesh Bank, police, communications sector, prosecutors, and courts. Thirdly, technological sufficiency requires that the law must allow for detection of cross-channel, networked and adaptive patterns, and that data must be integrated in a lawful manner. Finally, procedural sufficiency inquires if searches, freezes, blocking, confiscation and algorithmic evidence are authorized, documented and reviewable. Fifthly, the sufficiency question of rights-based surveillance and data processing is whether it is needed, proportionate, minimised, secure and contestable.^{^[23]}

The model suggests that low overall sufficiency and good offence coverage are possible. A statute can lawfully classify a behavior as criminal but fail to yield warranted or viable results because of a failure of institutions to share information, a failure of models to be audited, a failure of courts to be able to examine the automated evidence, and a failure of innocent individual to correct errors. However, rigorous protection does not mean ineffective enforcement, a risk ranking system that is open and comprehensible, a targeted approach to preservation, a process of human documentation and a swift judicial review can make enforcement more effective and more legitimate.^{^[24]}

Chapter 3: Methodology

3.1 Research Design

The research conducted in the study is of doctrinal type with qualitative approach, supplemented by policy and technology assessment. Doctrinal analysis refers to the analysis of the legal rules that govern online gambling, transfers of winnings, predicate offences, suspicious transaction reporting, financial intelligence, investigation, restraint, confiscation, blocking and surveillance. Afterwards, the rules are analyzed to determine if they establish a coherent enforcement path for digital-hundi and gambling-related laundering. It is not an evaluative approach (as opposed to empirical causal inference): it doesn't calculate the prevalence of gambling or the accuracy of a specific AI model.^{^[25]}

3.2 Materials

Primary materials include the Constitution of Bangladesh, the Gambling Prevention Act 2026, the MLPA 2012, Bangladesh Bank payment and foreign exchange directions and official BFIU materials. International benchmarks are based on FATF Recommendations, FATF guidance on new technologies and virtual assets and APG mutual evaluation of Bangladesh. Secondary materials consist of peer-reviewed online-gambling laundering, electronic smurfing, machine-learning-based AML, explainable AI and financial-surveillance governance research. The contemporary legal commentary should be used merely to identify concerns that are publicly expressed and not considered to be the law.^{^[26]}

3.3 Unit of Analysis and Case Typology.

The focus of the analysis is the legal provisions. They are looked at on a case-by-case basis and as part of an enforcement chain: detection; generation of suspicion; reporting and intelligence analysis; preservation or freezing; investigation and search; proof of the predicate offence and laundering elements; prosecution; confiscation; and review and remedy. The BFIU 'Z Casino' case study is developed on the basis of the new Act and features are directly relevant to the new Act and that is why it is used as a structured typology. The names listed in the official report are fake and the typology is not considered to be a case that has been proven in court.^{^[27]}

3.4 Statute-to-Standard Coding Matrix

The six following criteria are used as a code for each legal issue: (1) clarity of the legal basis; (2) the coverage of the risk; (3) designation of a competent authority; (4) availability of an operational mechanism; (5) procedural and rights safeguards; and (6) consistency with the MLPA and FATF architecture. A finding is categorized at ‘strong' level if the statute expressly addresses the conduct and provides an identifiable implementation pathway; ‘partial' if there is a coverage, but it relies on rules, institutional practice and/or an un-resolved interpretation; and ‘insufficient' if there is no material safeguard or operational element covered. The labels are not judicial statements, but instead analytical judgements.

The principal benchmark provisions are FATF Recommendations 1, 3, 4, 10, 15, 20, 29, 30, 31, 37 and 40. These are focused on Risk Assessment, Laundering Offences, Confiscation, Customer Due Diligence, New technologies and virtual assets, Suspicious transaction reporting, FIU functions, Investigative powers and Cooperation. Constitutional review is on articles: 31, 32, 39 and 43 in the context of article 18(2). The analysis considers whether restrictions are lawfully authorised and rationally related to the aim of preventing gambling and whether they incorporate safeguards that could prevent arbitrary interference.^{^[28]}

3.5 Evaluation of AI-Assisted Detection

The technological assessment does not rely on any particular model. It doesn't endorse any particular system, nor suggest that AI is more powerful than rules. Rather, it outlines legally relevant controls throughout the lifecycle of the model, from purpose definition to training and validation, performance testing, drift monitoring, explanation, human review, audit logs, adverse-action thresholds and challenge procedures. FATF guidance and peer-reviewed AML studies helps to differentiate between useful prioritisation and automated adjudication. ^{^[29]}

3.6 Reliability, Scope and Ethics

Source hierarchy and triangulation are used to support reliability. The legal description is regulated by official gazettes and statutory texts, institutional facts and typologies are substantiated by official reports and reports by the FIU, international standards serve as benchmarks, while the technological capabilities and limitations are explained by academic sources. The 2026 Act is new, so there is no reported case law on the provisions of the Act. The study does not, therefore, make predictions but rather expresses interpretative uncertainties.^{^[30]}

There are no human subjects, personal account records or confidential law enforcement information involved. The BFIU case is only discussed at the aggregate level that is publicized. While the analysis does not take forward a blanket call for the collection of personal data, it proposes that the notion of enforceability or data security and contestability are part of the privacy agenda, rather than separate barriers. Legislation and public documents are cut-off on 25 July 2026.^{^[31]}

Chapter 4: Results and Discussion

4.1 Substantive Offence Coverage: Strong

The 2026 Act is a very extensive piece of legislation regarding gambling. It acknowledges the virtual nature of the relevant 'place' and that gambling infrastructure comprises of applications, servers, domains, mirror sites, digital wallets, virtual assets, identity tools and payment accounts. There are no games to prove and no physical gaming house to show that there was no dependence in the case of online gambling offences and online betting offences. Other clauses cover bookmakers, match or spot fixing, advertising and promotion, facilitation via VPN or proxy infrastructure, fake SIM or MFS accounts, biometric fraud, facilitation and corporate participation. This breadth is directly related to the modularity of current gambling networks, where the ownership of a platform, collection of payments, promotions, technical hosting, and settlement could be handled by different actors.^{^[32]}

The money-laundering response is focused on Section 14. It makes the use of bank, MFS, digital wallet, hundi or any other financial system to transfer, conceal or legalise the fund for gambling, a crime and covers converting, storing or transfer by cryptocurrency, virtual currency or digital asset. The specific mention of ‘hundi’ is of legal significance as it indicates settlement arrangements that are not necessarily through an outward remittance. The express mention of virtual assets is also conducive to the FATF red-flag guidance's requirements for speed, pseudonymity and cross-border nature. ^{^[33]}

However, careful charging of the offence is required. The funds must be listed under 'Gambling funds' and mental elements must be proved and not presumed based on a flagged account. Payments can be accepted by both legitimate merchants and gambling players, devices can be used by families or agents, and VPNs can provide legitimate security and privacy. The statutory elements, knowledge or intention and evidence of facilitation purpose must be satisfied, and not a risk score alone.^{^[34]}

4.2 Integration with the MLPA: Strong but Interpretively Sensitive

Under section 25 of the MLPA, the section 14 conduct related to gambling-fund transactions, laundering and cryptocurrency use is a predicate offence. With the MLPA provision, access to BFIU reporting/analysis, temporary freezing, financial investigation, prosecution and confiscation are opened. The MLPA already includes provisions for conversion, movement, concealment, possession, use, structuring, participation and cross-border transfer of proceeds, and it expressly eliminates the requirement for a prior predicate-offence conviction. The unified structure can then advance the betting business and even subsequently attempt to conceal or take pleasure in its earnings.^{^[35]}

There is still a possibility of a overlap. Section 14 provides for the use of money-laundering terms and section 25 defines that behaviour as a predicate offence for MLP purposes. Where the same predicate is asserted in both charge 14 and on the laundering of the entire laundering act, it may be necessary for the court to decide whether the charges contain distinct predicate elements and facts. A coherent approach would recognize the gambling-fund offence as the creation, receipt or unlicensed receipt of gambling funds, and the MLPA offence as another or related concealment, conversion or structuring, possession or transfer with the necessary knowledge. By separating facts, duplication is eliminated and the principle of punishing the same act more than once is avoided.^{^[36]}

The reporting-organisation perimeter also has an impact. Bangladesh Bank can identify the additional institutions with government approval to be included in the MLPA, while banks, financial institutions and money remitters are included in the definition of MLPA. Because of this, payment-system and MFS actors will be in a better position to report suspicious patterns. The Act should be enacted in such a manner that the intelligence information gathered from gambling platforms is converted to MLPA-friendly narrative that includes customer identity, chronology of transactions, linked accounts, device-based hints, domain-based hints, reason for suspicion, supporting records, etc. A label like 'AI gambling score 92' alone would not be enough to assess BFIU meaningfully or subject the system to judicial review.^{^[37]}

4.3 Institutional and Enforcement Powers: Broad

The 2026 Act has a multi-layered approach to enforcement. Police officers can investigate crimes, courts can order freezing of bank, MFS, digital-wallet and cryptocurrency accounts, the Cyber Tribunal handles online and cyberspace crime and offences are cognisable, non-bailable, non-compoundable, and courts can confiscate money, property, accounts, wallets, servers, domains, SIMs and devices. Mirror or clone infrastructure is blocked as is the rest of the infrastructure (website, application, server, domain, IP and social media blocking powers expand to cover mirror or clone infrastructure). A national blacklist could include details of the offender, NID, SIM, MFS, bank, wallet, device, domain, IP etc. Parts 41-43 also call for identity linking, biometric or risk-based verification, inter-agency task force and delineation of the roles of the BTRC, Bangladesh Bank, BFIU and others. ^{^[38]}

These powers are suitable for the conditions of operation, where the illegal platforms move fast, and payment trails can be lost. Broad power doesn't necessarily mean institutional sufficiency, though. Duplicate freezing requests, unclear identifiers, late preservation, varying priorities of agencies and poor evidence turnover can jeopardize cases. The Act would be improved by a standard operating protocol that defines the authorities for each stage of an investigation, the verification of emergency indicators, the deconfliction of information, chain of custody and how an investigation moves from a gambling offence to an MLPA case.^{^[39]}

The APG's 2016 evaluation is a warning bell assessment. Bangladesh also had significant areas of financial intelligence and international cooperation, but financial intelligence for money-laundering investigation and prosecution was weak, and enforcement through confiscation and preventive measures was also underdeveloped, and risk-based supervision and predates investigations needed to be enhanced and expanded beyond corruption. The 2026 Act enhances legal instruments; however, the effectiveness of the Act will be subject to investigators, prosecutors, forensic capacity, and judicial practice. Having a blacklist or AI platform is no replacement for case development and asset-tracing skills.^{^[40]}

4.4 Section 44 and the Design of AI-Assisted Detection

Section 44 specifically allows for the use of AI monitoring systems, deep packet inspection, risk scoring, transaction monitoring, data analytics and other modern technologies. It also considers suspicious transactions, websites, applications, devices and wallets and accounts analysis. This is a good enabling clause as it allows for multi-source detection and would not permit enforcement solely on the basis of a specific database or fixed rule. The Bangladesh Bank circular also called on the payment entities to monitor the relevant payment merchants and customers through Artificial Intelligence and act accordingly if found. ^{^[41]}

Some sort of system that is both legally and technically sound should be operated in layers. Layer one would be based on other controls like a blocked domain, a known gambling merchant identifier, sanctioned wallet, impossible identity combinations and repeated use of fake or mismatched credentials. Layer two would have to apply behavioral elements like many-to-one collections, one-to-many dispersion, low balance retention, repetitive round value transfers, rapid cash-in and cash-out, unusual night-time velocity, and sudden changes in merchant volume, or transactions that don't match with the customer's occupation or merchant type. Layer three would be able to leverage graph analytics to discover common devices, beneficiaries, agents, settlement accounts, IP clusters/paths to virtual-asset off-ramps. Layer 4 would be alerts to be displayed to humans. The final determination of whether or not to file an STR, freeze an account or search the person or add an individual to a blacklist should be the responsibility of an authorised individual with reasons recorded.^{^[42]}

This is a layered structure with the same layers as are found in the research literature. If historical outcomes and contextual features are available, machine learning can be used to enhance prioritisation, though the effectiveness of the process is reliant on the quality of the labels, and the need for escalated and non-escalated cases. FATF acknowledges that new technologies can enhance risk identification while also alleviating repetitive work and that data quality and explainability are necessary, as are privacy and governance. Explainable models can give information about the contribution of features and relationships but the investigator must decide whether the features have case-specific meaningfulness and/or lawfulness. ^{^[43]}

4.5 Application to the BFIU 'Z Casino' Typology

The 'Z Casino' pattern would most likely produce multiple individual signals under a well-managed system. First, the collection pattern was many-to-many, which involved 51 accounts across 50 businesses, the bulk of which were repetitive low denomination collections that were not typical in retail. Secondly, the listed sectors by the companies such as e-commerce, telemedicine, construction, restaurants and advertising, failed to clarify a clear path of payments in relation to the gambling industry. Thirdly, money was funneled into 10 trust-cum-settlement accounts and then passed on to individuals whose profiles were not a match to the originator businesses. Fourthly, the proposed finalization via a hundi or hawala tied up with the domestic digital payment to an offshore obligation. ^{^[44]}

An AI system might rank this network up high, but without any one BDT 100 transfer being considered criminal. Centrality of graph nodes may be used to locate collection points; community detection may be used to find clusters of connected merchants; velocity and pass-through may be used to find settlement accounts; and merchant-profile models may be used to detect inconsistencies among merchants. Intelligence from domain, app, device and beneficiary may be corroborating. The legally relevant output should be an evidence package, including a transaction map, dates and amount, opening of an account and records of the KYC, merchant, device and login information (where lawfully obtained), links to the gambling platform, and reasons for the activity to have fallen from expected use. This package would enable an STR and a targeted investigation and would give the investigators the opportunity to test other hypotheses.^{^[45]}

The typology also demonstrates the need to consider other factors beyond identity matching as the basis for detection. They could be nominees, stolen biometrics, ghost SIMs or merchant fronts used by criminal networks. Linkage and biometric verification of identities provided in Section 41 could help to minimize identity fragmentation, but could also result in high impact errors if an identity is compromised and treated as evidence. Risk systems should thus be able to differentiate between the legal identity of an account and the suspected controller and should allow for correcting the situation if an innocent legal identity has been abused.^{^[46]}

4.6 Legal Sufficiency Matrix

Table 1. Legal Sufficiency Matrix

Dimension	Assessment	Reason
Substantive coverage	Strong	Online gambling, betting, fund transfers, hundi, virtual assets, facilitation, corporate liability and confiscation are expressly addressed.
Predicate-offence and MLPA linkage	Strong/partial	Section 25 activates the MLPA architecture, but section 14/MLPA overlap requires careful charging and interpretation.
Detection and intelligence powers	Strong	Section 44 and MLPA sections 23-25 enable AI monitoring, risk scoring, STR analysis and freezing.
Institutional coordination	Partial	Task-force and agency duties exist, but operational leadership, information standards and evidence handover require detailed protocols.
Algorithmic governance	Insufficient	No express standards for data quality, validation, drift, explainability, bias, audit logs, procurement or independent testing.
Privacy and proportionality	Partial/insufficient	Legitimate aims are clear, but DPI, blacklisting and urgent search powers lack detailed targeting, minimisation and oversight rules.
Procedural redress	Insufficient	The Act does not clearly specify notice, correction, appeal or compensation pathways for erroneous scores, blocking or blacklist entries.
Virtual-asset and cross-border response	Partial	Crypto and cooperation are recognised, but VASP supervision, travel-rule implementation and rapid foreign preservation remain operational challenges.
Evidential use of AI	Partial	AI may generate leads, but standards for disclosure, reproducibility, expert evidence and proof of statutory elements are not specified.

Source: Author’s doctrinal assessment of the 2026 Act, MLPA 2012, constitutional provisions and FATF standards.^{^[47]}

The matrix shows a decisive asymmetry. The new law is strongest where it defines offences and grants enforcement powers; it is weakest where automated suspicion produces consequences for individuals and businesses. This does not render the Act invalid. It means that section 48 rule-making, agency directions and judicial interpretation carry an unusually important burden.^{^[48]}

4.7 Privacy, Proportionality and Due Process

The goal of the Constitution is legitimate and expressed: The State is obliged to prevent gambling. The question is whether a particular monitoring or enforcement technique is reasonably connected and no broader than necessary. Article 43: privacy of correspondence and communication (with reasonable restrictions); articles 31 and 32: lawful treatment and personal liberty. Therefore, deep packet inspection is more sensitive than normal transaction monitoring, which is what Section 44 refers to. Transaction monitoring typically scans of regulated financial transactions for AML; DPI can scan network traffic that may contain legal communications that are not gambling related.^{^[49]}

The requirement to inspect content or population scale should be reserved for specific situations; in cases where it is not possible to inspect the content, it should be done using less intrusive metadata or targeted identifiers, and should be subject to judicial or independent authorisation for escalation. Should state purpose limitation, authorised users, access logging, encryption, retention and deletion. A recent academic article has rightly pointed out that there are no rules or explicit judicial supervision for DPI in s 44 and that the urgency of s 36 means that the power to search on the basis of volatility might be abused. The nexus principle, which was introduced by Oali Ahad, helps to establish a need that the facts documented by the officer will link the target, the data or the device to the specific threat of the gambling. ^{^[50]}

The Act also must have safeguards for account closure, blocking and blacklisting. While a temporary cover up may be warranted to not cause any frustration to an investigation, indefinite negative effects should not be based on an undisclosed model. The affected individual should at least be given a review mechanism, a statement of the non-sensitive reasons, be given a chance to demonstrate identity theft or legitimate business activity, and the maintained records should be corrected. Cascading exclusion is particularly important in the case where one list is propagated across NID, SIM, MFS, bank, wallet, domain and device systems where one error can lead to the next.^{^[51]}

4.8 Status of AI Outputs (evidential status)

An AI alert is an investigative lead or expert-assisted analytical output and not evidence of guilt. The prosecution is required to establish the statutory conduct and mental element by evidence that is admissible. In a model derived disclosure, information that may be required includes the purpose of the model, the version of the model, the material features of the model, the threshold, when the model was executed, data sources for the model, the validation results for the model, known limitations of the model, and the human action performed after the alert. A defendant or court should not be barred from challenging the basis of a coercive measure or a criminal allegation by proprietary confidentiality.^{^[52]}

Reproducibility will vary. With a deterministic rule, it can be re-run; a continuously updated machine-learning model can change after the event. So agencies need to maintain the model version, any parameters used, input snapshot, output and audit trail for each consequential decision. Explainability must be fit for purpose. A reason like 'high risk due to transaction velocity, shared beneficiary and link to blocked gambling domain' is more helpful than a generic confidence score. Each stated factor should then be confirmed with the involved records and witness/forensic evidence by the investigator.^{^[53]}

Enforcement quality is also safeguarded. If alerts are not documented, then they are weak alerts and it is challenging to learn from acquittals, false positives or emerging typologies. A feedback loop should be established to identify alerts that are closed, reported, investigated, prosecuted or referred to asset recovery, but not assume that non-prosecution means the alert was incorrect. Governance should monitor precision, recall if measurable, investigator workload, time to disposition, demographic/geographic disparities, and the percentage of high impact actions that were reversed.^{^[54]}

4.9 Virtual Assets and International Cooperation

While the Act's scope of cryptocurrency, digital assets and wallets is a significant positive, there is a need for implementation beyond criminalisation when it comes to virtual-asset tracing. FATF Recommendation 15 calls for mitigation of virtual-asset risks, identification and regulation or supervision of virtual-asset service providers, and FATF guidance and red flags concern over anonymity-enhancing services, rapid cross-border transfers and unusual transaction patterns and connections with high-risk services. Bangladesh requires a clear strategy for dealing with domestic VASPs, foreign exchanges that are accessible to Bangladeshi users, the ability to attribute wallets, preservation requests and the evidential treatment of blockchain analytics.^{^[55]}

The provisions of Section 46 allow for international and regional cooperation, extradition, exchange of information and recovery of assets. The MLPA also allows for FIU to FIU exchange and governmental agreements. These provisions are in line with FATF Recommendations 37-40, but there is a time constraint: a domain, wallet balance or exchange account can be shifted before formal mutual legal assistance has been completed. Operational arrangements should then be different from those undertaken for later requests for evidence, which may be for the purpose of preserving intelligence and information.Operational arrangements should therefore differentiate urgent intelligence sharing and preservation from later requests for evidence, which could be for the purpose of preserving intelligence and information. Requests shall include stable identifiers, transaction hashes, timing and restraining basis.^{^[56]}

4.10 Overall Finding

H1 is supported. The 2026 Act substantially fills the substantive void by introducing a new section on remote gambling and new sections on collection of funds and settlement of winnings through hundi, MFS and digital assets, and by associating the behaviour with the MLPA. H2 is also supported. The Act does not set out a full algorithmic-governance code, a privacy code or a redress code of its own but grants the power for surveillance and automated detection. Supported: The enforcement chain relies on coordinated data, the quality of STR, forensic investigation, prosecutorial theory, judicial oversight and cross-border cooperation.^{^[57]}

The legal framework thus can be characterized as having a strong content but still a less complete procedural and technological framework. It has the capacity to lawfully and effectively enforce with the aid of AI, where section 44 is seen as a direction to prioritize and investigate risk, not as a direction to automated punishment without review. The design of rules and institutional practices that will be implemented following the enactment will either shape the framework into a targeted AML approach or a general surveillance system that lacks strong evidence.^{^[58]}

Chapter 5: Policy Implications, Limitations and Conclusion

5.1 Policy Implications

Immediate policy priority is to introduce detailed rules or binding inter-agency standards in anticipation of high impact action at scale with section 44 tools. The instrument should specify allowable uses and sources of data, include provisions for the minimisation of data, include provisions for access of data based on roles, include provisions for encryption of data, include provisions for retention periods and deletion of data, and not allow for re-use for other purposes unless there is a new legal basis. DPI should be targeted and independently approved as much as possible in the narrowest of emergency situations, with ex post review as soon as possible.^{^[59]}

First, an algorithmic-governance framework; All models require an accountable owner, stated objective, data inventory, model feature rationale, model threshold approval, model drift monitoring and retirement procedure. Independent testing should include false-positive, false-negative, stability and subgroup effects and susceptibility to evasion. Any consequential actions such as filing an STR, extended freezing, blacklisting, blocking, search or arrest should be documented and require human review. The official should note the threshold (the legal one) and the supporting facts and not just approve a score. FATF's technology guidance enables innovation that's explainable, respects privacy, and is governed. ^{^[60]}

Thirdly, a gambling and digital-hundi reporting template should be issued by BFIU. It must include account roles, the chronology of transactions, MFS/bank affiliation, merchant category, device and domain information, suspected foreign settlement, and inconsistent account information with the customer profile. The template would facilitate the translation of reports from platforms or payment gateways into STRs that are MLPA-compliant, thereby minimizing the amount of unstructured narratives. Feedback to reporting organisations should be descriptive, without disclosing operationally sensitive intelligence, and should include useful typologies.^{^[61]}

Fourth, there must be a formal case-coordinating process that establishes evidence standards and the point person. BTRC and platform authorities reserve the right to retain domain and network data; payment providers retain customers and transaction records; BFIU analyse and share financial intelligence; police can obtain warrants and search; the distinction between the section 14 predicate offence and the laundering charge under BFIU; and court review of restraint, blocking and confiscation. Digital evidence should be documented, where applicable, be hashed, and have a statement of lawful acquisition.^{^[62]}

Fifthly, there needs to be a system of redress in place. Conditions for entering the digital blacklist and identity controls linked to it should be documented, reviews should be made at regular intervals, identity theft cases should be corrected, removal should be automatic after the expiration of legal action and there should be an independent decision-maker. Unless renewed, or on evidence, the freezing of accounts should be time-bound. Notice may be delayed if necessary to preserve an investigation but affected persons shall be provided with adequate reasons to contest an adverse action at some point thereafter. Such protections serve to strengthen rather than to weaken the constitutional prohibition on gambling in articles 31, 32 and 43. ^{^[63]}

Last but not least the arrangements for virtual assets and cross-border should be in keeping with recommendation 15 of the FATF and cooperation standards. Authorities require some quick preservation mechanisms for foreign exchanges and FIUs, and wallet and transaction-hash mechanisms, evidence for blockchain analytics, and a well-defined supervisory policy of VASPs for Bangladeshi customers. Annual reporting under section 47 should include overall totals for alerts, STRs, freezes, reversals, prosecutions, confiscation, model performance and complaints. Transparency would allow Parliament and courts, as well as the public, to evaluate effectiveness without revealing personal or operational data.^{^[64]}

5.2 Limitations

This study is not a comprehensive one because it was time-limited. The 2026 Act had only been in force for less than a month at the time of the studies and there was no judicial interpretation or implementation rules that were settled at the time of the studies. It is, therefore, legal design that is assessed in the analysis, not actual prosecution or model performance. The official Act is in Bangla, and English descriptions are functional translations and should be consulted with any authoritative English text which are published later by section 50. The case studies in public BFIU cannot be used to determine prevalence, causation or guilt and have been anonymised. The APG mutual evaluation is based on the situation as of 2015-16 and merely serves as a historical institutional benchmark. The implementation of rules under the Act, court rulings, enforcement results, asset recovery and independent audits of deployed systems should be studied.^{^[65]}

5.3 Conclusion

The Gambling Prevention Act 2026 is a significant reform of the gambling law in Bangladesh. Identifies online platforms, digital payment systems, hundi, cryptocurrency, identity abuse and network facilitation, enacts meaningful penalties and confiscation tools, and connects the conduct of gambling funds to the financial-intelligence and predicate-offence powers of the MLPA. Section 44 also provides a straightforward access path for authorities to leverage AI, risk scoring, transaction monitoring and analytics that will go beyond what fixed rules and manual review can achieve. ^{^[66]}

It is not the lack of power that constitutes the weak spot, it is the lack of detailed constraints on the adverse decisions that power can be used to make. AI systems can uncover settlement networks and repetitive low-value settlements, and highlight suspicious settlements. They can even create biased labels, misidentify valid sellers, and trigger a domino effect of account, communication and liberty repercussions. The Act does not yet provide sufficient detail on validation, explainability, minimisation of data, retention, independent audit, human review, notice, correction or appeal.^{^[67]}

The assessment is therefore overall one of partial sufficiency: strong from the perspective of substance and institutionally promising and technologically enabling, but still not completed from the perspective of proceedings. It can be effective and constitutionally sustainable when the automated output is investigative leads rather than coercive measures, the coercive measures are targeted and subject to review, the affected person's evidentiary pathway in the MLPA is well documented and the affected person has meaningful avenues for redress. If those protections are not in place, there could be more alerts, but weaker cases and more rights risk. The effectiveness of the new law will be judged not on the level of surveillance or account closures, but on the ability to accurately disrupt gambling connected laundering and fair adjudication and recover of criminal assets.^{^[68]}

Chapter 6: Bibliography

6.1 Legislation

The Constitution of the People's Republic of Bangladesh 1972

Gambling Prevention Act 2026 (Act No 98 of 2026), Bangladesh Gazette Extraordinary, 1 July 2026, 20497-20511

Money Laundering Prevention Act 2012 (Act No 5 of 2012)

Public Gambling Act 1867 (Act No II of 1867) (repealed)

6.2 Case Law

Oali Ahad v Government of the People's Republic of Bangladesh (1974) 26 DLR 376 (HCD)

6.3 Treaties and International Instruments

International Covenant on Civil and Political Rights (adopted 16 December 1966, entered into force 23 March 1976) 999 UNTS 171

6.4 Official Reports, Standards and Regulatory Materials

Asia/Pacific Group on Money Laundering, Anti-Money Laundering and Counter-Terrorist Financing Measures – Bangladesh: Mutual Evaluation Report (APG 2016) <https://www.fatf-gafi.org/en/publications/Mutualevaluations/Mer-bangladesh-2016.html> accessed 27 July 2026

Bangladesh Bank, Payment Systems Department, PSD Circular Letter No 04 (28 May 2025) <https://www.bb.org.bd/mediaroom/circulars/psd/may282025psdl04.pdf> accessed 27 July 2026

Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024) <https://www.bb.org.bd/pub/annual/bfiu/bfiu_2023-24.pdf> accessed 27 July 2026

Bangladesh Financial Intelligence Unit, Annual Report 2024-2025 (Bangladesh Bank 2026) <https://www.bfiu.org.bd/pdf/pub/annual/annualreport_2024_2025.pdf> accessed 27 July 2026

Bangladesh Financial Intelligence Unit, Guidance on Reporting Suspicious Transaction Report for the Reporting Organization <https://www.bfiu.org.bd/index.php/legislation/guidanceNote> accessed 27 July 2026

Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025) <https://www.fatf-gafi.org/en/publications/Fatfrecommendations/Fatf-recommendations.html> accessed 27 July 2026

Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021) <https://www.fatf-gafi.org/en/publications/Digitaltransformation/Opportunities-challenges-new-technologies-for-aml-cft.html> accessed 27 July 2026

Financial Action Task Force, Stocktake on Data Pooling, Collaborative Analytics and Data Protection (FATF 2021) <https://www.fatf-gafi.org/en/publications/Digitaltransformation/Data-pooling-collaborative-analytics-data-protection.html> accessed 27 July 2026

Financial Action Task Force, Updated Guidance for a Risk-Based Approach to Virtual Assets and Virtual Asset Service Providers (FATF 2021) <https://www.fatf-gafi.org/en/publications/Fatfrecommendations/Guidance-rba-virtual-assets-2021.html> accessed 27 July 2026

Financial Action Task Force, Virtual Assets Red Flag Indicators of Money Laundering and Terrorist Financing (FATF 2020) <https://www.fatf-gafi.org/en/publications/Methodsandtrends/Virtual-assets-red-flag-indicators.html> accessed 27 July 2026

6.5 Journal Articles and Research Papers

Brooks G, 'Online Gambling and Money Laundering: “Views from the Inside”' (2012) 15(3) Journal of Money Laundering Control 304 <https://doi.org/10.1108/13685201211238052> accessed 27 July 2026

Fiedler I, 'Online Gambling as a Game Changer to Money Laundering?' (2013) SSRN Electronic Journal <https://doi.org/10.2139/ssrn.2261266> accessed 27 July 2026

Jensen RIT and Iosifidis A, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889 <https://doi.org/10.1109/ACCESS.2023.3239549> accessed 27 July 2026

Jullum M, Loland A, Huseby RB, Anonsen G and Lorentzen J, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173 <https://doi.org/10.1108/JMLC-07-2019-0055> accessed 27 July 2026

Narasapuram NK, Shah SAA, Shiratuddin MF and Sohel F, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics 111 <https://doi.org/10.1007/s13042-026-02994-w> accessed 27 July 2026

Thommandru A, 'Smurfing in Electronic Banking: A Legal Investigation of the Potential for Transnational Money Laundering' (2023) 51(1) International Journal of Legal Information 69 <https://doi.org/10.1017/jli.2023.13> accessed 27 July 2026

Vilone G and Longo L, 'Notions of Explainability and Evaluation Approaches for Explainable Artificial Intelligence' (2021) 76 Information Fusion 89 <https://doi.org/10.1016/j.inffus.2021.05.009> accessed 27 July 2026

6.6 News and Online Commentary

Aysharja DAA, 'Could the New Gambling Law Threaten Citizens’ Privacy?' The Daily Star (20 July 2026) <https://www.thedailystar.net/opinion/views/news/could-the-new-gambling-law-threaten-citizens-privacy-4227721> accessed 27 July 2026

BSS, 'Bangladesh Enacts New Anti-Online Gambling Law' The Business Standard (2 July 2026) <https://www.tbsnews.net/bangladesh/bangladesh-enacts-new-anti-online-gambling-law-1477991> accessed 27 July 2026

Star Online Report, 'Parliament Passes Anti-Gambling Law; up to 7 Years' Jail, Tk 5cr Fine for Online Betting' The Daily Star (30 June 2026) <https://www.thedailystar.net/news/the-parliament-watch/parliament/news/parliament-passes-anti-gambling-law-7-years-jail-tk-5cr-fine-online-betting-4212331> accessed 27 July 2026

^{^[1]} Constitution of the People's Republic of Bangladesh 1972, art 18(2); Public Gambling Act 1867 (Act No II of 1867) (repealed).

^{^[2]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 2, 9-14, 25 and 44.

^{^[3]} I Fiedler, 'Online Gambling as a Game Changer to Money Laundering?' (2013) SSRN Electronic Journal, doi:10.2139/ssrn.2261266; A Thommandru, 'Smurfing in Electronic Banking: A Legal Investigation of the Potential for Transnational Money Laundering' (2023) 51(1) International Journal of Legal Information 69, doi:10.1017/jli.2023.13; Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024).

^{^[4]} Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024).

^{^[5]} Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024); Bangladesh Financial Intelligence Unit, Annual Report 2024-2025 (Bangladesh Bank 2026).

^{^[6]} Gambling Prevention Act 2026 (Act No 98 of 2026) s 44; Bangladesh Bank, Payment Systems Department, PSD Circular Letter No 04 (28 May 2025); Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021).

^{^[7]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 36 and 40-44; Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 art 43; Oali Ahad v Government of the People's Republic of Bangladesh (1974) 26 DLR 376 (HCD).

^{^[8]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 14, 25, 36 and 40-44; Money Laundering Prevention Act 2012 (Act No 5 of 2012) ss 23-25.

^{^[9]} Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025); Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 arts 31, 32 and 43.

^{^[10]} I Fiedler, 'Online Gambling as a Game Changer to Money Laundering?' (2013) SSRN Electronic Journal, doi:10.2139/ssrn.2261266; G Brooks, 'Online Gambling and Money Laundering: Views from the Inside' (2012) 15(3) Journal of Money Laundering Control 304, doi:10.1108/13685201211238052.

^{^[11]} A Thommandru, 'Smurfing in Electronic Banking: A Legal Investigation of the Potential for Transnational Money Laundering' (2023) 51(1) International Journal of Legal Information 69, doi:10.1017/jli.2023.13; Financial Action Task Force, Virtual Assets Red Flag Indicators of Money Laundering and Terrorist Financing (FATF 2020).

^{^[12]} Money Laundering Prevention Act 2012 (Act No 5 of 2012) ss 2 and 4; A Thommandru, 'Smurfing in Electronic Banking: A Legal Investigation of the Potential for Transnational Money Laundering' (2023) 51(1) International Journal of Legal Information 69, doi:10.1017/jli.2023.13.

^{^[13]} Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024); M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549.

^{^[14]} Money Laundering Prevention Act 2012 (Act No 5 of 2012) ss 2, 4 and 23-25; Bangladesh Financial Intelligence Unit, Guidance on Reporting Suspicious Transaction Report for the Reporting Organization.

^{^[15]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 14 and 25; Money Laundering Prevention Act 2012 (Act No 5 of 2012) s 4.

^{^[16]} Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025) Recommendations 1, 3, 4, 10, 15, 20, 29-31 and 37-40.

^{^[17]} M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549.

^{^[18]} Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549.

^{^[19]} RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549; M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w.

^{^[20]} NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w; G Vilone and L Longo, 'Notions of Explainability and Evaluation Approaches for Explainable Artificial Intelligence' (2021) 76 Information Fusion 89, doi:10.1016/j.inffus.2021.05.009.

^{^[21]} Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 art 43; International Covenant on Civil and Political Rights (adopted 16 December 1966, entered into force 23 March 1976) 999 UNTS 171, art 17; Oali Ahad v Government of the People's Republic of Bangladesh (1974) 26 DLR 376 (HCD).

^{^[22]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 36 and 44; DAA Aysharja, 'Could the New Gambling Law Threaten Citizens' Privacy?' The Daily Star (20 July 2026); Star Online Report, 'Parliament Passes Anti-Gambling Law; up to 7 Years' Jail, Tk 5cr Fine for Online Betting' The Daily Star (30 June 2026).

^{^[23]} Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025); Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 arts 31, 32 and 43.

^{^[24]} Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); Financial Action Task Force, Stocktake on Data Pooling, Collaborative Analytics and Data Protection (FATF 2021); G Vilone and L Longo, 'Notions of Explainability and Evaluation Approaches for Explainable Artificial Intelligence' (2021) 76 Information Fusion 89, doi:10.1016/j.inffus.2021.05.009.

^{^[25]} Gambling Prevention Act 2026 (Act No 98 of 2026); Money Laundering Prevention Act 2012 (Act No 5 of 2012); Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025).

^{^[26]} Gambling Prevention Act 2026 (Act No 98 of 2026); Money Laundering Prevention Act 2012 (Act No 5 of 2012); Bangladesh Bank, Payment Systems Department, PSD Circular Letter No 04 (28 May 2025); Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024); Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025); Asia/Pacific Group on Money Laundering, Anti-Money Laundering and Counter-Terrorist Financing Measures - Bangladesh: Mutual Evaluation Report (APG 2016).

^{^[27]} Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024).

^{^[28]} Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025) Recommendations 1, 3, 4, 10, 15, 20, 29-31 and 37-40; Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 arts 18(2), 31, 32, 39 and 43.

^{^[29]} Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549; NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w; G Vilone and L Longo, 'Notions of Explainability and Evaluation Approaches for Explainable Artificial Intelligence' (2021) 76 Information Fusion 89, doi:10.1016/j.inffus.2021.05.009.

^{^[30]} Gambling Prevention Act 2026 (Act No 98 of 2026); Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024); Asia/Pacific Group on Money Laundering, Anti-Money Laundering and Counter-Terrorist Financing Measures - Bangladesh: Mutual Evaluation Report (APG 2016).

^{^[31]} Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024); Financial Action Task Force, Stocktake on Data Pooling, Collaborative Analytics and Data Protection (FATF 2021).

^{^[32]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 2-13, 17 and 20-30; BSS, 'Bangladesh Enacts New Anti-Online Gambling Law' The Business Standard (2 July 2026); Star Online Report, 'Parliament Passes Anti-Gambling Law; up to 7 Years' Jail, Tk 5cr Fine for Online Betting' The Daily Star (30 June 2026).

^{^[33]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 14 and 25; Financial Action Task Force, Virtual Assets Red Flag Indicators of Money Laundering and Terrorist Financing (FATF 2020).

^{^[34]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 14 and 25; Money Laundering Prevention Act 2012 (Act No 5 of 2012) s 4.

^{^[35]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 14 and 25; Money Laundering Prevention Act 2012 (Act No 5 of 2012) ss 4 and 23-25.

^{^[36]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 14 and 25; Money Laundering Prevention Act 2012 (Act No 5 of 2012) s 4.

^{^[37]} Money Laundering Prevention Act 2012 (Act No 5 of 2012) ss 2 and 23-25; Bangladesh Financial Intelligence Unit, Guidance on Reporting Suspicious Transaction Report for the Reporting Organization.

^{^[38]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 28-46; BSS, 'Bangladesh Enacts New Anti-Online Gambling Law' The Business Standard (2 July 2026).

^{^[39]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 36 and 40-46; Asia/Pacific Group on Money Laundering, Anti-Money Laundering and Counter-Terrorist Financing Measures - Bangladesh: Mutual Evaluation Report (APG 2016).

^{^[40]} Asia/Pacific Group on Money Laundering, Anti-Money Laundering and Counter-Terrorist Financing Measures - Bangladesh: Mutual Evaluation Report (APG 2016).

^{^[41]} Gambling Prevention Act 2026 (Act No 98 of 2026) s 44; Bangladesh Bank, Payment Systems Department, PSD Circular Letter No 04 (28 May 2025); Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021).

^{^[42]} M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549; Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w.

^{^[43]} M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549; Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w.

^{^[44]} Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024).

^{^[45]} Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024); M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549.

^{^[46]} Gambling Prevention Act 2026 (Act No 98 of 2026) s 41; Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024).

^{^[47]} Gambling Prevention Act 2026 (Act No 98 of 2026); Money Laundering Prevention Act 2012 (Act No 5 of 2012); Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43; Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025).

^{^[48]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 44 and 48; Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 arts 31, 32 and 43.

^{^[49]} Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 arts 18(2), 31, 32 and 43; Gambling Prevention Act 2026 (Act No 98 of 2026) s 44.

^{^[50]} DAA Aysharja, 'Could the New Gambling Law Threaten Citizens' Privacy?' The Daily Star (20 July 2026); International Covenant on Civil and Political Rights (adopted 16 December 1966, entered into force 23 March 1976) 999 UNTS 171, art 17; Oali Ahad v Government of the People's Republic of Bangladesh (1974) 26 DLR 376 (HCD).

^{^[51]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 40-43; Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 arts 31, 32 and 43.

^{^[52]} RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549; M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055; NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w; G Vilone and L Longo, 'Notions of Explainability and Evaluation Approaches for Explainable Artificial Intelligence' (2021) 76 Information Fusion 89, doi:10.1016/j.inffus.2021.05.009; Money Laundering Prevention Act 2012 (Act No 5 of 2012) s 4.

^{^[53]} NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w; G Vilone and L Longo, 'Notions of Explainability and Evaluation Approaches for Explainable Artificial Intelligence' (2021) 76 Information Fusion 89, doi:10.1016/j.inffus.2021.05.009.

^{^[54]} Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); RIT Jensen and A Iosifidis, 'Fighting Money Laundering with Statistics and Machine Learning' (2023) 11 IEEE Access 8889, doi:10.1109/ACCESS.2023.3239549; M Jullum and others, 'Detecting Money Laundering Transactions with Machine Learning' (2020) 23(1) Journal of Money Laundering Control 173, doi:10.1108/JMLC-07-2019-0055.

^{^[55]} Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025) Recommendation 15; Financial Action Task Force, Updated Guidance for a Risk-Based Approach to Virtual Assets and Virtual Asset Service Providers (FATF 2021); Financial Action Task Force, Virtual Assets Red Flag Indicators of Money Laundering and Terrorist Financing (FATF 2020).

^{^[56]} Gambling Prevention Act 2026 (Act No 98 of 2026) s 46; Money Laundering Prevention Act 2012 (Act No 5 of 2012) s 24(4); Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025) Recommendations 37-40.

^{^[57]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 14, 25 and 44; Money Laundering Prevention Act 2012 (Act No 5 of 2012) ss 4 and 23-25.

^{^[58]} Gambling Prevention Act 2026 (Act No 98 of 2026) s 44; Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); DAA Aysharja, 'Could the New Gambling Law Threaten Citizens' Privacy?' The Daily Star (20 July 2026).

^{^[59]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 36, 44 and 48; Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 art 43; International Covenant on Civil and Political Rights (adopted 16 December 1966, entered into force 23 March 1976) 999 UNTS 171, art 17.

^{^[60]} Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); Financial Action Task Force, Stocktake on Data Pooling, Collaborative Analytics and Data Protection (FATF 2021); NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w; G Vilone and L Longo, 'Notions of Explainability and Evaluation Approaches for Explainable Artificial Intelligence' (2021) 76 Information Fusion 89, doi:10.1016/j.inffus.2021.05.009.

^{^[61]} Bangladesh Financial Intelligence Unit, Guidance on Reporting Suspicious Transaction Report for the Reporting Organization; Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024).

^{^[62]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 14, 25, 36 and 42-46; Money Laundering Prevention Act 2012 (Act No 5 of 2012) ss 23-25.

^{^[63]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 40-43; Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43 arts 31, 32 and 43.

^{^[64]} Financial Action Task Force, International Standards on Combating Money Laundering and the Financing of Terrorism and Proliferation: The FATF Recommendations (FATF 2012, amended October 2025) Recommendations 15 and 37-40; Financial Action Task Force, Updated Guidance for a Risk-Based Approach to Virtual Assets and Virtual Asset Service Providers (FATF 2021); Gambling Prevention Act 2026 (Act No 98 of 2026) ss 46-47.

^{^[65]} Gambling Prevention Act 2026 (Act No 98 of 2026) s 50; Bangladesh Financial Intelligence Unit, Annual Report 2023-2024 (Bangladesh Bank 2024); Asia/Pacific Group on Money Laundering, Anti-Money Laundering and Counter-Terrorist Financing Measures - Bangladesh: Mutual Evaluation Report (APG 2016).

^{^[66]} Gambling Prevention Act 2026 (Act No 98 of 2026) ss 2, 14, 25 and 44; Money Laundering Prevention Act 2012 (Act No 5 of 2012) ss 4 and 23-25.

^{^[67]} Gambling Prevention Act 2026 (Act No 98 of 2026) s 44; Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021); Financial Action Task Force, Stocktake on Data Pooling, Collaborative Analytics and Data Protection (FATF 2021); NK Narasapuram and others, 'Explainable Artificial Intelligence Models for Detecting Suspicious Bank Transactions' (2026) 17 International Journal of Machine Learning and Cybernetics art 111, doi:10.1007/s13042-026-02994-w; G Vilone and L Longo, 'Notions of Explainability and Evaluation Approaches for Explainable Artificial Intelligence' (2021) 76 Information Fusion 89, doi:10.1016/j.inffus.2021.05.009; DAA Aysharja, 'Could the New Gambling Law Threaten Citizens' Privacy?' The Daily Star (20 July 2026).

^{^[68]} Gambling Prevention Act 2026 (Act No 98 of 2026); Money Laundering Prevention Act 2012 (Act No 5 of 2012); Constitution of the People's Republic of Bangladesh 1972, arts 18(2), 31, 32, 39 and 43; Financial Action Task Force, Opportunities and Challenges of New Technologies for AML/CFT (FATF 2021).

How to Cite This Article

NAZIA RAHMAN, MOHAMMAD MINHAZUR RAHMAN, AI-Assisted Detection of Digital Hundi and Online Gambling-Linked Money Laundering: Evaluating the Legal Sufficiency of Bangladesh's New Gambling Prevention Act 2026 and MLPA 2012., White Black Legal – International Law Journal, ISSN: 2581-8503, Vol. Volume 4, Issue VOLUME 4 ISSUE 2, October 2026, pp. 1278-1310. Available at: https://www.whiteblacklegal.co.in/details/ai-assisted-detection-of-digital-hundi-and-online-gambling-linked-money-laundering-evaluating-the-legal-sufficiency-of-bangladeshs-new-gambling-prevention-act-2026-and-mlpa-2012

Author & Publication Record

Authors: NAZIA RAHMAN & MOHAMMAD MINHAZUR RAHMAN

Registration ID: 107208 | Published Paper ID: WBL7208 & WBL7209

Year: Oct- 2026 | Volume: 4 | Issue: 2

Approved ISSN: 2581-8503 | Country: Delhi, India

Page No.: 1278-1310

Full Text Preview Open in New Tab

Article Information

Journal White Black Legal

ISSN 2581-8503

VolumeVolume 4

IssueVOLUME 4 ISSUE 2

PublishedOctober 2026

Pages1278-1310

Registration ID107208

Paper IDWBL7208 & WBL7209Year: Oct- 2026

CountryDelhi, India Page No.: 1278-1310

AccessOpen Access

Article Tools

Download Full Paper PDF Search on Google Scholar Search on Google

Copied