Developing Google's Ad Map for Prediction Markets Now Has Five Holes In It. The Holes Follow the Cease-and-Desists, Not the Court Wins

Editor's note: TrueEdge builds odds tools and earns affiliate commissions from licensed sportsbooks. This piece argues a position on prediction-market regulation. Weigh it accordingly.

The notice is two sentences long. It is titled "Update to Prediction Markets (October 2026)," it was posted to Google's advertising policy change log on October 7, and it says that in October 2026 Google will prohibit the advertisement of prediction-market contracts and related products in Connecticut, effective October 7. It gives no reason. It does not say what a related product is. It does not mention the Department of Consumer Protection, the nine cease-and-desist orders of September 10, the nearly 30 subpoenas that went out the same day, or Judge Vernon Oliver's August ruling that Kalshi's sports contracts "are sports wagers." It did not need to. As of this weekend Google's policy page lists one approved location for prediction-market advertising, and it reads "United States (excluding Connecticut, Michigan, Nevada, New York, and Ohio)."

Five states in nine months. The question is what the five have in common, because it is not what the exchanges would like it to be.

The map

Nevada was excluded from the day the policy took effect, January 21; by then the state had already beaten Crypto.com's exchange in its own district court, and on August 28 the Ninth Circuit would uphold Nevada's position against Kalshi, Crypto.com and Robinhood. Ohio was added June 2, the day a federal court there denied Kalshi a preliminary injunction. Michigan and New York were added July 13, per Gambling Insider's reporting as carried by PPC Land; Michigan has no court ruling against the exchanges that we are aware of, and New York's suit against Polymarket came later, in September. Connecticut was added October 7, two months after Oliver's ruling and four weeks after the state's orders.

So the list is not a list of states that have won in court. Nevada and Connecticut have; Ohio won at the district level in June and at the Sixth Circuit in September; Michigan and New York have letters and a lawsuit. What the five share is an enforcement posture: a regulator that has put something in writing to the platforms and, in most cases, to the companies around them. Google's policy page says advertisers must "comply with all local laws," and a two-sentence notice is what that clause looks like when a state attorney general or consumer-protection commissioner has said, in writing, that the product is illegal there. Google is not adjudicating the Commodity Exchange Act. It is reading its mail.

That is the whole point, and Connecticut understood it before anyone.

What Connecticut built in September

The orders of September 10 went to nine companies: Polymarket, Coinbase, Crypto.com, Robinhood, ProphetX, Novig, Webull, Gemini and Underdog Predict. They demanded that each stop advertising, offering or promoting sports event contracts to Connecticut residents and let customers withdraw their balances. That part is ordinary; several states had done something like it by then. The part that was not ordinary was the second envelope. The Department of Consumer Protection issued nearly 30 subpoenas the same day, to payment processors, data providers, and, per PPC Land's count, 15 media organisations. WFSB, the CBS affiliate in Hartford, reported receiving one. The state's line was that the recipients were not under investigation. They were being told, formally, that the state regarded the product they were carrying, processing or promoting as unlicensed gambling.

This site wrote on September 12 that Connecticut had stopped trying to win the preemption fight and subpoenaed Apple instead. Five weeks later the state has the one thing a subpoena to a media company cannot deliver on its own, which is a platform acting on its own account. Google was not subpoenaed, as far as we know. It did not have to be. On October 7 the Department of Consumer Protection told CT Insider that three of the nine, ProphetX, Gemini and Webull, had withdrawn from the state. An analyst disputed the ProphetX claim and the company did not respond to the trade press, so count it as two confirmed and one contested. The same day, Google's notice appeared. We do not know whether the state asked for it. The notice does not say, and Google has not said.

Commissioner Bryan Cafferelli's argument in September was about youth: one platform, he said, reported nearly $250 million in college-football trading volume on the first day of the season, and "the dollars show the youth is clearly at risk." Whatever one thinks of that argument, it is not one a federal court has to accept for it to work on an advertising network. An ad policy is not a preemption analysis. It is a risk decision by a company with no stake in the outcome and every reason to avoid being the next subpoena.

The strongest case against reading it this way

The best argument on the other side has two parts, and the first is simply scale. Kalshi did $75.4 billion in trades over the thirty days to mid-October, per Aldrin Research, with nearly $4 billion of it on a single NFL Sunday. Nobody built that on Google search ads in Connecticut. Prediction markets grew on CNN and CNBC partnerships, on sports-radio integrations, on an app that shows up on a phone without being searched for. Losing paid search in five states is a line item, not a strategy problem, and the exchanges can still target the other forty-five with a federal licence Google explicitly recognises. The policy page itself says who may advertise: a CFTC-designated contract market or an NFA-registered broker. That is a federal definition sitting at the top of the policy. The states got five footnotes.

The second part is that Connecticut did not need any of this. It won in court. Oliver held in August that the contracts are not swaps and that even if they were, "sports wagering has long been subject to state regulation pursuant to the states' police powers." A state with a district-court ruling in hand is not routing around the law. It is enforcing it, and the subpoenas and the ad exclusion are what enforcement of a court win looks like against companies that keep operating anyway.

We think the first argument is right about the money and wrong about the mechanism. The exchanges' position in every brief is that a federally licensed market cannot be regulated by fifty states, and this column has agreed with it: we prefer a federal rule to state criminal enforcement, and we have said so for months. But the distribution layer does not care about the brief. Google's exclusions are a fifty-state map by construction, drawn one notice at a time, and each notice makes the next cheaper. Michigan got on the list with no ruling at all. If the mechanism is "a regulator wrote a letter," then the exchanges' strongest legal argument, Supremacy, has no purchase on it, because nobody is being asked to decide whether the letter is right.

The second argument proves our point rather than theirs. Connecticut had a win in August and it still sent 30 subpoenas in September, because an injunction against Kalshi is an injunction against Kalshi. The subpoenas reached the companies that are not parties to anything: the processors, the data vendors, the stations that run the ads. That is not what a state does when it thinks the court win is enough. It is what a state does when it has noticed that the exchanges treat court losses as venue problems and file somewhere else.

What the CFTC's chairman said, and why it does not reach Mountain View

Michael Selig went on Fox Business on October 15 and said the thing the exchanges want said: "you can't offer an order book in a casino." The proposed rule his agency issued on October 9 would define sports event contracts as swaps and, in a companion interim rule, exclude sportsbook bets from the definition. If that rule survives, the exchanges' federal argument gets stronger in every circuit that has not yet ruled.

It changes nothing on Google's policy page. The page already accepts that the advertisers are federally licensed; that is the entry condition. What it adds is "comply with all local laws," and a CFTC rule does not amend Connecticut's gambling statute or withdraw a commissioner's order. The exchanges' theory is that the rule will eventually make those orders unenforceable. Perhaps. Until a court says so, Google will keep reading its mail, and the list on the policy page will keep getting longer in the direction of whichever state writes next.

What we do not know

We do not know whether Connecticut asked Google for the exclusion or whether Google acted on the public orders alone; the notice is silent and neither party has said. We do not know whether Apple, which received the state's subpoena in September, has changed anything about app distribution in Connecticut, which would matter far more than search ads. We do not know whether ProphetX is in the state or out of it. And we do not know whether Google's main policy page, which as of October 11 still listed only four excluded states per PPC Land and now lists five, will be the document the exchanges eventually challenge, or whether there is anything to challenge in a private company declining to sell them something.

If the Supreme Court takes New Jersey's petition and rules for the exchanges next year, every one of these exclusions becomes a policy Google can reverse in two sentences, the same way it wrote them. If it does not, the map on the policy page is the map, and the exchanges will have spent two years winning the argument about which agency regulates them while losing the argument about who can see their ads. The second one is the one that decides where the customers come from.