Investigation Shows Chatbots Often Point Toward Offshore Casinos

Investigation Shows Chatbots Often Point Toward Offshore Casinos

Artificial intelligence chatbots have become a common tool for answering online questions, from travel planning to financial advice. A recent investigation suggests that when these systems are asked about online gambling, the responses may lead users toward operators that function outside national regulatory frameworks. The findings have drawn attention from policymakers and researchers studying the intersection of emerging technology and regulated industries.

The investigation, conducted by the journalism network Investigate Europe, examined how several well-known AI chatbots respond to questions about online casinos. The researchers tested systems developed by companies such as OpenAI, Google, and Meta Platforms. Over two weeks, the team ran the same set of prompts across 10 European countries to observe how the systems handled requests related to casino bonuses and accessing gambling platforms outside national restrictions.

Across those tests, the chatbots frequently referenced offshore gambling sites that do not hold licences in the jurisdictions where the questions were asked. The responses sometimes described features commonly used in promotional material for such platforms, including fast withdrawal processes, minimal identity checks, and promotional incentives. Some of the websites mentioned in the chatbot answers appear on official blacklists maintained by European regulators.

The research involved a structured approach. Each chatbot received thirty prompts in every country included in the study. The prompts were designed to resemble questions that ordinary users might ask when searching for gambling information online. Some queries focused on identifying casinos that offer promotional bonuses, while others asked how individuals could gamble online after joining national self-exclusion programmes.

The results showed a consistent pattern. According to the investigation, Meta’s chatbot referred to unlicensed operators in twenty-seven of the tested responses. Google’s Gemini system produced similar recommendations in twenty-six cases, while ChatGPT referenced such sites twenty-two times. Although the numbers varied across countries, the pattern appeared in each jurisdiction included in the study.

The researchers also examined the geographical differences in the responses. In both France and Poland, the tests produced eighteen recommendations for offshore casinos. In the United Kingdom, fourteen references were recorded. The results suggest that the issue does not relate to a single national market but instead reflects the way generative AI systems gather and summarise information from publicly available sources.

Some chatbot responses echoed the language used in affiliate marketing content for offshore gambling operators. The investigation found that these explanations sometimes highlighted anonymity or reduced verification procedures as potential advantages. In one case cited in the report, a chatbot referred to so-called “no-ID casinos” using language that portrayed them as a notable option for users seeking fewer restrictions.

Questions about self-exclusion programmes produced similar outcomes. Self-exclusion tools are widely used across Europe as part of responsible gambling policies. They allow individuals to restrict their own access to licensed gambling platforms for a specific period of time. The United Kingdom operates one of the most widely known systems through Gamstop, a national programme that links participating gambling operators to a shared database of self-excluded users.

When the investigators asked the chatbots how individuals might continue gambling after registering with such programmes, the systems often suggested casinos that operate beyond the national licensing framework and therefore do not participate in those schemes. The responses sometimes included references to offshore operators that accept customers from multiple countries without applying the same identity verification requirements.

Cryptocurrency-based casinos also appeared regularly in the chatbot answers. These platforms were described as offering payment methods that do not rely on traditional banking systems. In promotional materials online, digital currencies are sometimes presented as a way to reduce identity checks or financial monitoring. The investigation suggests that generative AI systems may repeat such descriptions when summarising information from online sources.

According to the researchers, much of the material cited by the chatbots appeared to originate from marketing texts published by offshore gambling operators and affiliate websites. These sources are widely distributed across the web and often present casinos based in jurisdictions such as Curaçao as accessible alternatives to nationally regulated platforms.

The report frames the issue as a “routing problem.” In this context, routing refers to the possibility that users who ask AI assistants for guidance may be directed toward services operating outside established consumer protection frameworks. Traditional search engines have gradually introduced systems designed to limit the visibility of blacklisted gambling websites. Generative AI tools, however, do not simply display links. Instead, they generate direct written answers that summarise information from multiple sources.

This difference in format may influence how users interpret the information. When an AI chatbot presents a written explanation that includes references to offshore casinos, the response may appear more like advice than a list of search results. For regulators who have focused on advertising restrictions and search engine cooperation, the emergence of conversational AI introduces a new layer of complexity.

Concerns about the issue have been voiced by policymakers and organisations involved in gambling harm prevention. They argue that AI systems designed to assist with everyday queries could inadvertently reduce the effectiveness of consumer protection measures if they repeat promotional information from unlicensed operators.

Among those commenting on the findings is Tiemo Wölken, a member of the European Parliament who contributed to discussions surrounding the EU’s digital regulation framework. Speaking about the investigation, Wölken noted that AI chatbots can function as an enhanced form of search engine while lacking some of the safeguards that have been gradually introduced into traditional search platforms.

His remarks refer in part to the regulatory environment created by the Digital Services Act, a European Union law that aims to strengthen accountability among large online platforms. The legislation requires companies to address systemic risks linked to the distribution of illegal or harmful content. As generative AI tools become integrated into everyday digital services, policymakers may examine how those rules apply to automated conversational systems.

For technology companies, the investigation raises questions about how AI models process information related to regulated sectors. Developers rely on large collections of online text to train their systems. When those sources contain marketing material for offshore gambling sites, the language may reappear in chatbot responses unless additional filtering mechanisms are applied.