A Dutch study found that 10 leading generative AI tools named unlicensed online gambling operators in 33 of 79 interactions, with 3 tools implying legality. The study, conducted by XY Legal Solutions, raises concerns about AI recommendations undermining Dutch gambling regulations.
TL;DR
- 10 AI tools named unlicensed casinos in 33 of 79 interactions in a Dutch study.
- 3 tools implied legality, with Microsoft Copilot generating a false 'Top 50 legal casinos' list.
- The findings come as VNLOK sues Meta for illegal gambling ads on Facebook and Instagram.
What happened
The study, conducted by XY Legal Solutions between August and September 2026, tested 10 AI platforms using Dutch-language prompts. The platforms evaluated were ChatGPT, Google AI Mode and Gemini, Microsoft Copilot, Perplexity, Anthropic Claude, xAI Grok, DeepSeek, Mistral Le Chat, and Meta AI.
All 10 tools identified at least one non-licensed operator by name or domain in at least one conversation. Eight tools did so without any research disclaimer. Four tools — Mistral Le Chat, Google AI Mode, DeepSeek, and Grok — supplied such names after a single prompt.
Three assistants — Microsoft Copilot, Mistral Le Chat, and Google Gemini — presented unlicensed operators in ways that implied legality or licensing. Microsoft Copilot generated a 'Top 50 legal casinos' list containing unlicensed brands and erroneously linked them to the Kansspelautoriteit (KSA) and Cruks.
Why it matters
The findings highlight potential regulatory gaps in how AI tools handle unlicensed gambling promotion under Dutch rules. AI recommendations can undermine KSA licensing and Cruks controls, diverting traffic from compliant operators.
Regulators may need to implement targeted measures to address generative AI tools and platform ads. The study's results could influence future regulations and policies governing AI's role in promoting gambling services.
For developers and startups, the study underscores the importance of implementing robust content moderation and compliance measures in AI tools. Investors should consider the regulatory risks associated with AI tools that may inadvertently promote illegal activities.
Key facts
- The study tested 10 AI platforms: ChatGPT, Google AI Mode and Gemini, Microsoft Copilot, Perplexity, Anthropic Claude, xAI Grok, DeepSeek, Mistral Le Chat, and Meta AI.
- All 10 tools named at least one unlicensed operator in at least one conversation.
- 33 of 79 interactions involved the AI tools naming unlicensed operators.
- Four tools named unlicensed operators after a single prompt: Mistral Le Chat, Google AI Mode, DeepSeek, and Grok.
- Three tools implied legality: Microsoft Copilot, Mistral Le Chat, and Google Gemini.
- Microsoft Copilot generated a false 'Top 50 legal casinos' list containing unlicensed brands.
- VNLOK is suing Meta for illegal gambling ads on Facebook and Instagram.
- The study was conducted by XY Legal Solutions between August and September 2026.
Context
The findings come as VNLOK, a Dutch trade association, pursues Meta in court for alleged illegal gambling advertisements on Facebook and Instagram. The case highlights the ongoing challenge of insufficient cooperation from Meta on illegal gambling ads.
The study's results could influence future regulations and policies governing AI's role in promoting gambling services. It also underscores the need for developers and startups to implement robust content moderation and compliance measures in AI tools.
The broader AI landscape is increasingly facing scrutiny over its role in promoting illegal or unlicensed activities. This study adds to the growing body of research highlighting the need for regulatory oversight and compliance in AI tools.