.svg)
The EU’s landmark legislation on artificial intelligence (AI) has already partially come into effect, raising new compliance challenges for a gambling industry that has been eagerly integrating AI into many facets of its operations.
Gambling’s rush to cram AI into its products now faces new challenges if any of their business activities touch EU consumers.
In August of last year, the European Union’s new high-level legislation governing the use of AI came into force, with provisions gradually coming into effect through to August 2027.
The legislation “is designed to ensure that AI developed and used in the EU is trustworthy, with safeguards to protect people's fundamental rights”, according to the European Commission.
The positive news for the gambling industry is that any AI system deployed in a gambling context is unlikely to fall under the scope of the EU’s “high-risk” classification.
High-risk systems come with the heaviest compliance burden and even though a key consideration for riskiness under the act is user “safety”, which could be compromised by a faulty responsible gambling AI, there are other qualifying factors which gambling does not meet.
For example, high-risk systems must relate in some way to an area in which the EU has harmonised legislation.
Infamously, gambling has never been the subject of lawmaking at the bloc level and remains one of the most devolved areas of law in the European Union.
Pain Points
More likely to be relevant to the gambling industry are the eight “prohibited practices” laid out by the AI Act, which have been banned since February 2 this year across all AI systems.
These include the use of AI to develop deceptive or manipulative techniques, which raises concerns for the use of AI in marketing or to alter the way that games behave in response to how people gamble.
Several major online operators are already touting their use of AI to deliver custom marketing promotions and to personalise the customer journey.
Dubious marketing practices have been central to numerous gambling enforcement cases over the past decade and enmeshing advertising and AI creates obvious compliance risks beyond how they may intersect with the EU’s new law.
A ban on “social scoring” is also a possible area of risk for safer gambling tools.
The practice is defined by the European Commission as “evaluating or classifying individuals or groups based on social behaviour or personal traits, causing detrimental or unfavourable treatment of those people”.
But its broad definition has come in for criticism and there are fears that it might capture a wide range of business practices.
Financial services professionals, for example, are concerned that the established use of advanced AI systems to predict when individuals are about to commit fraud might now be illegal in the EU.
The growing trend for regulators to favour affordability-style systems that consider a wide range of evidence about an individual to assess if they are potentially vulnerable will also need to be carefully calibrated to ensure they do not engage in prohibited social scoring.
Fines for breaching any of these AI prohibitions can rise as high as €35m, or 7 percent of total worldwide annual turnover for the preceding financial year, whichever is higher.
Penalties can be applied all along the supply chain of a particular AI system, meaning that companies are in the firing line regardless of the various categories defined by the AI Act that a company might fall into, whether they created the system, are just making it available to consumers or anything in between.
The list is also subject to change, with the EU saying it will review and update the categories once a year.
Legal experts have noted that many uses of these eight otherwise forbidden practices can be navigated via exceptions to the basic text of the law.
Safer gambling needs may well fall within this remit, but as with many areas of compliance, companies will need to be prepared to justify their decision-making should EU inspectors come knocking.
At least one major operator that touts its use of AI to support its safer gambling practices says it is not worried.
A spokesperson for Entain told Vixio GamblingCompliance: “We have been monitoring the Act closely for some time. Our current assessment is that the Act will not have a material change to our ARC player protection system. However, we will continue to closely monitor the roll-out of the Act and any implications it may have for [Entain].”
Let’s Chat
Numerous gambling companies now use generative AI to power the first line of their customer support and the act introduces several requirements for AI-powered chatbots.
In many cases it can be unclear to consumers whether or not they are interacting with a real human being, but from August 2, 2026, users must be made aware if they are speaking to AI.
The method of making this clear is left up to the provider, but must take into account the message’s target audience and ensure it meets any accessibility needs they might reasonably be expected to have, the law says.
Any content generated by AI must also be clearly marked once the law comes fully into effect next August.
Article 50(2) states: “Providers of AI systems ... generating synthetic audio, image, video or text content, shall ensure that the outputs of the AI system are marked in a machine-readable format and detectable as artificially generated or manipulated.”
