MGA publishes voluntary AI Gaming Charter for responsible AI use
The Malta Gaming Authority (MGA) has published a voluntary charter providing guidance on the responsible use of artificial intelligence across the gaming sector.
According to the MGA, the Charter follows research findings that identified gaps in governance and risk management when it comes to AI in the gambling sector.
The MGA said that its AI Gaming Charter does not create new legal or regulatory obligations but instead sets out best-practice guidance for operators using AI. Existing rules, including the EU AI Act and General Data Protection Regulation (GDPR), will still apply.
The Charter covers several gaming functions where AI is increasingly being used, including responsible gambling, anti-money laundering, fraud prevention, customer service, player profiling, payments and game development.
According to the MGA, the Charter takes a risk-based approach, with the level of safeguards depending on the potential impact of AI on players, their rights, gaming integrity and other related outcomes.
The Charter also calls for human oversight and greater transparency around AI systems, alongside measures covering fairness, data protection, security, reliability and accountability.
It encourages licensees to establish governance structures, carry out risk assessments, and monitor their AI systems, as well as set up processes to respond to incidents.
According to the MGA, industry research it commissioned showed that AI adoption was uneven among gaming operators; finding gaps in AI governance, risk assessment, transparency and bias testing, as well as the potential impact of AI on vulnerable or at-risk players.
The MGA said it developed the Charter with the Malta Digital Innovation Authority (MDIA) following engagement with licensees and other stakeholders and is encouraging all gaming licensees to adopt the framework voluntarily.
The Charter comes alongside growing scrutiny of how AI is used to analyze player behavior, with a recent New York Times investigation alleging that gambling operator DraftKings used AI to identify users more likely to continue gambling and lose money in response to promotions.
Verticals:
Sectors:
Dig Deeper
Related stories
The Backstory
AI moves from experiment to regulatory issue
The Malta Gaming Authority’s voluntary AI Gaming Charter lands at a point when artificial intelligence has shifted from a back-office efficiency tool to a central question for gambling regulators. Operators are using AI across customer service, fraud monitoring, anti-money laundering checks, payment risk, game design, marketing and responsible gambling interventions. That breadth makes the technology useful but also harder to supervise.
The MGA’s decision to publish guidance rather than impose new rules reflects a wider regulatory challenge: AI is already embedded in gaming operations, but many markets still lack sector-specific standards for how it should be governed. Existing laws, including the EU AI Act and the General Data Protection Regulation, set broad obligations. The gaming-specific question is how those obligations apply when automated systems assess player risk, personalize offers or flag possible harm.
The charter’s emphasis on human oversight, transparency, risk assessment and accountability is therefore less a theoretical exercise than a response to live operational pressure. Regulators are trying to balance the benefits of AI-driven monitoring with the risk that the same systems can be used to intensify engagement among vulnerable customers.
Governance gaps were already visible
The MGA’s approach closely tracks concerns highlighted in recent industry research. A University of Nevada, Las Vegas and KPMG study found that while AI adoption is now widespread across gaming companies, governance remains underdeveloped. The report, covered in COMPLETE iGAMING’s report on gaps in igaming AI governance, scored governance at just 30 out of 100 on its AI Maturity Index.
That finding matters because the report said about 80% of gaming companies had implemented some form of AI, while fewer than 20% had dedicated AI governance roles. It also found that 15% had no AI oversight policies, more than 42% had no plans to hire AI-specific staff and only 2% had embedded policies for responsible AI use. In other words, the industry’s use of the technology is racing ahead of the internal structures needed to manage it.
The report described a disconnect between companies’ stated concerns over privacy and governance and their willingness to fund the people and systems required to address those concerns. For regulators, that creates a practical risk. If operators lack clear ownership for AI systems, it becomes harder to identify who is responsible when models produce biased outcomes, misclassify customers or fail to detect harmful play.
The MGA charter appears designed to close that gap without immediately adding a new compliance layer. By encouraging licensees to create governance structures, conduct risk assessments and monitor AI systems, Malta is signaling what it expects mature operators to be able to demonstrate, even if the charter is voluntary.
Personalization sharpens the responsible gambling debate
The most sensitive area for AI in gambling is not automation itself but personalization. Operators can use machine learning to identify fraud, spot abnormal betting patterns and detect signs of distress. They can also use the same data infrastructure to refine promotions, segment players and predict who is likely to respond to incentives.
That dual use explains why responsible gambling features prominently in the MGA charter. A risk-based approach requires stronger safeguards when automated systems could affect player rights, gaming integrity or harm-prevention outcomes. The higher the potential impact on a customer, the more important transparency, human review and model testing become.
Recent scrutiny of promotional practices has intensified that debate. The MGA referred to growing concern about how AI is used to analyze player behavior, including a New York Times investigation that alleged DraftKings used AI to identify customers more likely to keep gambling and lose money in response to promotions. The allegation goes to the heart of the policy issue: predictive analytics can support intervention, but it can also make gambling products more efficient at extracting losses.
That tension will likely shape how regulators interpret responsible AI. If AI tools are used to identify at-risk customers, regulators may ask whether the operator used that information to restrict exposure, reduce marketing or trigger support. If the same indicators are used to target offers, operators could face a much tougher regulatory and reputational challenge.
Emerging markets add another layer of risk
The governance question is especially important as regulated online gambling expands into newer markets. The Philippines has become a prominent example of how quickly digital gambling can grow once licensing frameworks open. Suppliers are also finding that online customers may behave differently from traditional casino patrons, complicating assumptions about product design and risk controls.
Light & Wonder’s expansion experience, described in COMPLETE iGAMING’s interview on new regulated markets, showed that the Philippine market did not simply mirror land-based casino behavior. The company found a split between existing casino customers and remote online players who had little connection to casino brands. That lesson has broader relevance for AI oversight: models trained on one customer base may not behave reliably in another.
In fast-growing digital markets, regulators also face political pressure over access, advertising and social harm. In the Philippines, Cardinal Pablo David, president of the Catholic Bishops’ Conference, accused PAGCOR of promoting online gambling and called for a total ban, according to COMPLETE iGAMING’s coverage of the criticism of PAGCOR. His comments reflected a broader concern that online platforms can reach younger audiences and people far from physical casinos.
Lawmakers have raised similar concerns about promotion. Sen. Erwin Tulfo called for tighter scrutiny of gambling advertising, warning that illegal operators continued to advertise despite existing rules, as reported in COMPLETE iGAMING’s article on the push to curb online gambling ads. He has also filed a bill that would ban online gambling operations, betting, advertising and related financial transactions.
Those disputes show why AI governance cannot be separated from market structure. Where illegal operators advertise on social platforms and licensed operators use sophisticated personalization, regulators must distinguish between legitimate harm-prevention technology and tools that may accelerate risky play.
Regulators are asserting boundaries in adjacent markets
The MGA’s charter is part of a wider pattern of regulators clarifying boundaries as technology reshapes gambling products. In Nevada, the Gaming Control Board recently warned licensees that sports event contracts are wagers under state law, even when listed on federally regulated exchanges. COMPLETE iGAMING reported on Nevada’s warning that sports event contracts require sports-pool approval, including the risk that licensees could face disciplinary action for unauthorized activity or problematic partnerships.
The Nevada notice was not about AI, but it reflects the same regulatory instinct. New products often develop faster than traditional gaming statutes, leaving regulators to reassert jurisdiction after operators and adjacent financial platforms test the perimeter. AI presents a similar boundary problem because it can sit inside licensed systems without appearing as a separate gambling product.
For Malta, the voluntary charter gives the regulator a way to set expectations before enforcement cases define the field. It also gives operators a benchmark against which investors, partners and future regulators may judge them. Even without creating new legal duties, the charter may influence what is considered reasonable conduct when AI is used in player profiling, payments, fraud detection or responsible gambling.
The stakes are commercial as well as regulatory. Companies that build credible governance early may be better placed to expand across markets that are beginning to scrutinize digital gambling more closely. Those that rely on opaque models, thin oversight and aggressive personalization risk finding that voluntary guidance becomes the template for future mandatory rules.
