AI Responsible-Gambling Tools Compared: GameScanner, mentor, BetBuddy, ARC

Every large operator now says it uses data to spot players at risk of harm, and a handful of specialist suppliers sell machine-learning tools to do it. This guide compares the ones that are named in public documents with a named user, using only what those documents say. Each tool has a full profile in our AI products tracker.

All four take behavioural play data (stakes, deposits, session length, loss chasing) and produce a risk score or profile with the reasons behind it; each product page lists the exact inputs and outputs its sources describe.

GameScanner (Mindway AI)

Mindway AI describes GameScanner as combining neuroscience research with expert assessments to detect at-risk and problem gambling automatically, and to explain why each player was flagged. Its partners include Interwetten, Midnite and Crown Resorts. Mindway calls the Crown deal the first land-based application in Australia. Mindway says the tool works as a “virtual psychologist” that detects at least 87% of the problem-gambling cases a human expert would, and that its products track 16.5 million active players in 73 jurisdictions. Both figures are Mindway’s own. See also our story on Mindway’s GameScanner approach.

mentor (Neccton, owned by OpenBet)

Neccton, bought by OpenBet in 2023, scores player risk in real time, alerts compliance teams and sends players personalised feedback. Finland’s Veikkaus and Fanatics Sportsbook are named users. mentor has the strongest published evidence of any tool here: a 2015 peer-reviewed study in Frontiers in Psychology found that 1,015 gamblers who received personalised feedback spent significantly less time and money than 15,216 matched controls. The study was co-authored by Neccton’s managing director, which is why we rate it a named deployment rather than independent evidence.

BetBuddy (Playtech)

Playtech bought BetBuddy in 2017. It compares each player’s activity with patterns of known risky gamblers, and Playtech says it uses “explainable AI”: it shows the behavioural markers behind each score, not just the score. A 2016 paper in International Gambling Studies on predicting self-exclusion with supervised machine learning was co-authored by BetBuddy’s chief executive.

ARC (Entain)

ARC is Entain’s in-house player-protection programme. In February 2021, when it completed ARC’s first phase, Entain said’s models use more than three times the number of behavioural indicators it used before, including stake fluctuations, erratic single-session play and loss chasing. Entain said it works with Harvard Medical School’s Division on Addiction, but we found no published independent evaluation.

Operators building their own

Large operators also run their own models, and their SEC filings say so:

• FanDuel: Flutter’s Q2 2025 results release says FanDuel’s “Real-Time Check-In feature uses machine learning to detect risk and generate personalized interventions at the point of play.” Flutter also says its AceAI betting assistant points customers to help and stops showing bets when a prompt suggests a problem.
• MGM: MGM’s 2025 annual report says that “within the framework of LeoSafePlay, we have launched a tool based on machine learning and algorithms that help in the creation of risk profiles for customers who are at risk of developing a gaming problem.”

Each company’s AI statements are on its earnings page: Flutter, MGM.

What the evidence does and doesn’t show

None of these tools has a published evaluation that is both peer-reviewed and independent of the vendor. Detection rates are self-reported, and they measure agreement with a human expert or with later self-exclusion, not whether the intervention that followed reduced harm. When choosing a tool, an operator should ask for the validation data, what the model is trained on, and how staff act on a flag.

What regulators say

• Malta: The MGA’s AI Gaming Charter, launched on 18 September 2026, sets voluntary principles of transparency, fairness, data governance and human oversight. It creates no new legal obligations.
• Great Britain: In June 2026 the Gambling Commission’s director of enforcement said it is not against operators using AI for anti-money-laundering, but the evidence so far shows such tools too often fail to deliver what is required.
• Massachusetts: The rules restrict AI rather than require it. Operators may not use AI they expect to make a platform more addictive to target wagers or offers, and must report every six months on how they use patron data for responsible gaming.

For tools that catch fraud and match-fixing rather than harm, see AI fraud and integrity detection in gambling.