On 17 June 2026 the gambling authorities of Belgium, France, Germany, Italy, the Netherlands, Poland, Portugal, Spain and Switzerland published a joint statement on prediction markets on the website of the French Autorité nationale des jeux. It names no company, creates no new law, and made a fraction of the noise of any American court filing this summer. It is still the most consequential document the category has faced outside the United States, because it is the moment nine separate national answers stopped looking like a coincidence and started looking like a position.
The difference is not volume, it is the question. The American argument is about who gets to decide, and it is genuinely unresolved. That argument never started in Europe, because gambling is a national competence. The European question is narrower and much harder to argue your way out of: what is this product, and can a person sitting in our country reach it? Every authority that has answered so far has answered the same way, and the test they used is one you can run on yourself.
Nine authorities put their names on one page
The joint statement is short and reads like a list of things its signatories already agreed on. The risks it names are product properties rather than legal characterisations: continuous availability, no built in stake limits, no time limits, weak age and identity verification, exposure to trading on inside information, and volatility carried by retail participants.
The commitments are equally concrete. The authorities undertake to share information across borders, to monitor advertising and betting integrity in a coordinated way, and to escalate through warnings, blocking, fines, advertising restrictions and the freezing of accounts. The German Gemeinsame Glücksspielbehörde der Länder confirmed its participation on 19 June 2026, calling cross border exchange of information a precondition for acting against unlicensed offers.
One paragraph was aimed at nobody in the industry. The signatories told sports organisations that before entering sponsorship or commercial relationships they "doivent s'assurer de la légalité de ces plateformes", that they must satisfy themselves the platforms are lawful in that territory. The Belgian Commission des jeux de hasard published the same declaration and put it bluntly: prediction markets comply with none of the national rules and offer consumers no protection.
The timing was not subtle. The statement landed as the 2026 World Cup began, when football volumes rise and sponsorships are most visible. As iGaming Business reported, the coordination was the message.
There is no European passport, and that changes what compliance means
An American reader can miss how different the ground is. There is no European Union gambling licence, and no mutual recognition regime turning an authorisation in one member state into permission in another. Nine bodies applying nine different statutes still reached the same classification within about eighteen months, because the product looks the same from each vantage point.
Germany shows the sharpest version. The German position is not that an operator needs a licence it failed to obtain. It is that this category of wager, treated as a Gesellschaftswette, is not licensable at all. The GGL has warned consumers directly and added Polymarket to the German blocking list in September 2025.
Directive 05 of our manifest was written for exactly this: "Standards are set against the jurisdiction that regulates next, not only the one that has not yet." Europe is not the jurisdiction that regulates next. It is the one that already did, quietly, while attention was elsewhere. We wrote about the first European blocks in March 2026, when the pattern was visible but not yet settled.
The Dutch file shows what the test actually measures
The Netherlands produced the most legible record, because the Dutch Kansspelautoriteit publishes its decisions. Its case file on Adventure One QSS Inc., the company behind Polymarket, carries an enforcement order dated 20 January 2026, a collection decision dated 19 May 2026 and a decision on objection dated 23 June 2026.
Read the findings and notice what is missing. When the KSA announced the order on 17 February 2026, its reasoning rested on what an investigator could observe. A user with a Dutch IP address could reach the site, create an account, deposit money, take a position and withdraw funds. Alongside that the authority listed indicators of orientation toward Dutch consumers: automated support answering in Dutch, payments in euros, and markets on Dutch political events. The penalty was set at 420,000 euros per week, capped at 840,000 euros.
Not one of those findings is a legal conclusion about the nature of an event contract. Every one is an observable property of a live product. The European determination does not wait for a court to characterise your instrument; it is made by someone opening your website. Which means you can make it first, for every country you can be reached from, and write it down with a date on it.
One day of rollout time cost 420,000 euros
The detail worth the most to an operator sits in the collection decision. Polymarket had until 17 February 2026 to stop serving Dutch residents. According to CasinoNieuws.nl, which read the file, the block went live on 18 February 2026. One day past the deadline, and the 420,000 euro penalty fell due automatically. The company argued the blocks had been deployed but needed time to propagate. The regulator's answer was that the deadline had passed. The penalty went unpaid, producing the collection decision of 19 May 2026, and the objection was rejected on 23 June 2026.
Strip away the company and the lesson is uncomfortable for most venues. Geofencing is not a switch. It is a deployment, with propagation time, cached sessions and payment rails that do not all stop at the same instant, and almost nobody has measured their own. If a regulator gave you fourteen days to become unreachable from one country, what is your elapsed time from decision to verified block, and have you tested that number rather than estimated it? It is the discipline we argued for when the CFTC ordered an exchange to keep operating in August 2026, where continuity is engineered in advance, not granted at the moment of crisis. Access control is the same problem pointed the other way.
France escalated for twenty months before it pulled the switch
The French sequence answers anyone who believes these decisions arrive without warning. The ANJ's attention was drawn to the service in November 2024 and a formal notice followed. The company geoblocked financial transactions from French territory, a measure the authority later called easy to circumvent. In February 2026 the ANJ said publicly that prediction markets have no legal existence in France without authorisation.
The platform kept growing anyway. In its blocking decision of 16 July 2026, the ANJ recorded 578,751 visits and 205,057 unique visitors from France in June 2026 alone, and noted the absence of any system for identifying users. The president of the authority then ordered French internet service providers to block the site.
Two things there travel well beyond France. A partial block that was trivially circumvented was treated as no block at all, because effectiveness was measured by whether French users actually reached the service, not by whether a control had been implemented in good faith. And the homepage itself was the problem: displaying live odds was characterised as advertising an unauthorised offer, an offence the authority ties to a fine of 100,000 euros. That lands on precisely the surface the CFTC's staff wrote about in August 2026 when they said a price display is a compliance signal rather than a design choice. Two legal systems with nothing in common arrived at the same place.
Portugal shows the integrity case arriving through the election door
Portugal moved fastest, and for a different reason. Betting on political events is not among the permitted categories under the Portuguese framework, which covers sports betting, horse racing and games of chance, so the classification was straightforward. What made it urgent was the trading pattern around the presidential first round of 18 January 2026.
The regulator, the Serviço de Regulação e Inspeção de Jogos, notified the operator on 19 January 2026 to cease activity. As ECO reported that day, the site was declared illegal while remaining reachable for Portuguese users. Blocking orders followed, and Renascença reported on 21 January 2026 that MEO, Vodafone and NOS had blocked access.
The volume figures deserve a note on how we treat numbers. ECO put election related turnover at roughly 90 million euros; CoinDesk reported above 103 million euros. Neither comes from an audited source, so both are reported rather than established. When the most cited numbers about a market's activity are reconstructions by outsiders who disagree, the venue has left the description of its own scale to other people. Directive 02 asks for the opposite.
What escalated Portugal was not the licensing question, which had been true for years. It was a pattern around an official result that looked, to a regulator, like something other than forecasting. That is the fastest route from a quiet legal problem to an urgent one, and it is why we treat unfair informational advantage as the category's trust test.
Spain blocked a federally regulated exchange, and that is the part to sit with
On 26 May 2026 the Spanish Ministerio de Derechos Sociales, Consumo y Agenda 2030 opened sanction proceedings and ordered precautionary blocking of two platforms: Polymarket and Kalshi. The grounds were operating without Spanish authorisation and the absence of identity verification, age controls, controls for self excluded persons and adequate supervision. The regulator, the Dirección General de Ordenación del Juego, had been unable to reach the companies at their foreign addresses.
Kalshi is a designated contract market regulated by the CFTC, and that status is the centre of its legal position in the United States. In Spain it was neither a defence nor the question. Spanish law asks whether you hold Spanish authorisation, and federal recognition in a third country is not an input to that test.
This is not a criticism of the company. It is a structural fact about European gambling regulation, and the most useful thing an American operator can take from the file: regulatory achievement at home is not portable. The sentence "we are federally regulated" answers a question Madrid, The Hague and Paris are not asking. It mirrors what we described when two American courts split over the same statute, where the defensible posture is the strictest live reading.
A Gibraltar licence did not settle the German question
The declaration's paragraph on sponsorship stopped being theoretical within a day. On 17 June 2026 Sportschau reported that the GGL had opened a review into ADI Predictstreet, the official prediction markets partner of the 2026 World Cup. The company holds a Gibraltar licence obtained on 8 April 2026 and no German licence, while its branding appeared on stadium signage and in tournament broadcasts, a setting where only providers on the German whitelist may advertise. A spokesperson, quoted in the same reporting, said no breach occurs from the company's perspective and that it directs no marketing at Germany specifically.
Two structural points do the work. A licence issued by one European jurisdiction did not answer the question in another, which is the passport problem again, inside Europe. And the defence that no marketing is directed at a country is what the Dutch authority declined to accept, having assembled observable indicators of orientation instead. The perimeter reaches past the product to sponsorships, affiliates and partners. If a tournament partnership can put a company inside a national advertising prohibition, partner selection is a licensing decision wearing a marketing budget.
What a venue should be able to answer about every country it reaches
None of these decisions required inside knowledge to anticipate. Each turned on facts a venue already holds about itself. Here is what we would want written down, dated and owned by a named person, for every jurisdiction reached.
- A reachability record you produced yourself. From a residential connection in each country: can someone register, fund an account, take a position and withdraw? Not a policy statement but an observation, repeated on a schedule and dated. This is the Dutch test verbatim.
- An orientation inventory. Languages support answers in, including automated support. Currencies and payment rails accepted. Whether the book carries markets about that country's politics or domestic competitions. The KSA treated all of these as evidence of targeting, whatever the terms said.
- A measured block time. Elapsed hours from internal decision to verified block, tested rather than estimated, including payment rails and cached sessions. The Dutch file prices a twenty four hour overrun at 420,000 euros.
- An advertising perimeter map. Sponsorships, affiliates, influencer arrangements and the front page itself. France counted an odds display as advertising and Germany counted stadium signage, both outside what most venues treat as their compliance boundary.
- A plan for positions that outlive access. What happens to residents' open positions when a country closes, written into the resolution terms beforehand rather than negotiated under a deadline.
- An escalation history you keep yourself. Every warning, notice and informal contact from every authority, in one place with dates. France's file ran twenty months, and a venue that cannot reconstruct its own timeline cannot argue about it.
None of it requires a legal opinion, and all of it is obtainable by a regulator without your cooperation. That asymmetry is the argument for producing it first.
The standard you write yourself is the one that travels
It would be easy to read the European file as hostility toward the category, and to answer that these authorities do not understand what an event contract is. That reading is comfortable and it is a trap, because it predicts nothing. It did not predict Belgium in early 2025, Portugal in January 2026, or Spain reaching a CFTC regulated exchange in May, and it says nothing about who moves next.
The more useful reading is that nine authorities in nine legal systems converged on the same test, and that the test is behavioural rather than doctrinal. Can a person here reach it, deposit into it, and lose money on it without meaningful identification, limits or protection? Every decision above answers that first and applies whatever local statute fits second. That is good news for anyone building this properly, because a behavioural test is one you pass by building rather than by arguing. A venue holding a dated reachability record, a measured block time, an orientation inventory and an advertising perimeter is holding the evidence any regulator eventually asks for, including the one it already answers to.
Our manifest puts it in one line, and the European file is the best demonstration of it so far: "The market that survives regulation is the one that was already behaving as if regulation had arrived." In nine countries, it already has.