Is AI threatening our anti-money laundering systems?
•
10 Oct 2025 •
Reading time: 3 minutes Artificial intelligence is now at the heart of the gambling industry’s compliance mechanisms. But behind the promise of efficiency and automation, the UKGC is sounding the alarm: these technologies, poorly managed or misunderstood, could undermine the fight against money laundering and the protection of players.
The alarm bells are ringing: AI is going beyond its role as a servant
The story begins in the hushed corridors of the UK Gambling Commission. During a recent review of its supervisory missions, the authority observed a worrying trend: the rapid rise of artificial intelligence tools in anti-money laundering (AML) systems, without operators fully understanding how they work. According to the regulator, some algorithms, poorly configured or misunderstood, fail to identify high-risk signals or do not trigger any follow-up, allowing certain murky activities to slip through the net.
The challenge goes beyond iGaming. At a time when sophisticated AI tools are being democratised, technological self-regulation is an illusion: AI acts like a distorting mirror, sometimes amplifying the risks it was supposed to help control.
Unsuspected risks
According to the Gambling Commission, some AI systems carry a hidden requirement: a minimum threshold of volume or duration of activity before generating a risk score. As a result, a customer can act intensively without triggering any alert at the start of his relationship with the operator.
In other cases, the algorithm escalated an account for review, but the managers didn’t know why. The compliance team was unable to identify what the model had detected. Such grey areas undermine operational accountability and make it difficult to trace decisions.
The Gambling Commission has also pointed to risky practices: some companies delegate the design of their tools to external service providers, sometimes without sufficient control. In several cases, incorrect information or information that did not comply with legislation was included in official documents.
New criminal tactics: deepfakes, crypto and mule accounts
AI, hijacked by malicious actors, sometimes becomes an accomplice to fraud. The UK regulator warns that AI programmes are now capable of producing forged documents, video deepfakes or artificially generated faces to circumvent identity checks.
In this context, online gaming operators need to train their teams to identify signs of artificiality in customer documents. The influx of crypto-assets is also exacerbating the situation: stolen wallets could be laundered via betting platforms, particularly through crash games where the average retention time is very short, making it more difficult to detect suspicious behaviour.
Finally, there is another worrying pattern: mule accounts. Certain profiles are recruited to create or add to customer accounts in exchange for payment, so that funds can be moved between entities without arousing suspicion.
Reactions and outlook: between caution and regulation
Faced with these emerging threats, the Gambling Commission is taking a proactive stance. It has announced that it will audit the AI tools used by operators to assess their relevance, compliance and effectiveness.
Operators will have to justify each threshold, weighting or trigger, and demonstrate that they have the in-house skills to understand these models. The days of black boxes are over: transparency is becoming a regulatory imperative.
In addition, the regulator points out that the licence conditions launched by the Gambling Commission (in particular condition 12.1.1(3)) require operators to adapt their compliance policies to emerging developments in crime.
A lesson for all sectors: AI between opportunity and danger
The warning issued by the Gambling Commission does not just concern the online gaming sector. It poses a central dilemma for all industries: how far can critical functions (financial, compliance, security) be delegated to AI without rigorous human oversight?
The emerging response is a need for re-humanisation: human teams must remain capable of deciphering algorithmic decisions, auditing models and intervening manually in case of doubt.
The rise of artificial intelligence tools, while opening up promising prospects, also exposes the flaws of a world where technological opacity can serve the interests of fraud. The UK Gambling Commission has sounded the alarm: operators who use ill-conceived or poorly understood algorithms run a real risk of non-compliance and severe penalties.