The headline numbers

Sportsbooks: AI feature status, operator by operator

The 10 licensed US operators are ranked by AI citation visibility (5W Research, 2026) — how often each is surfaced when people ask ChatGPT, Claude, Perplexity or Google AI Overviews about sports betting. Confirmed live AI feature shown in green; no confirmed live feature in red.

The industry’s biggest operator doesn’t have one

The most counter-intuitive result in this census is DraftKings. It is the sportsbook most Americans would name first, and along with FanDuel it captures roughly 68% of US sports betting handle (Track360 industry data) — yet as of this writing it has no live, publicly available AI chat or picks assistant. DraftKings filed a U.S. patent application for an AI betting assistant, but the application was rejected in May 2026, and the company has not announced a public launch date for a replacement. Industry coverage describes DraftKings as racing to build something that can compete with FanDuel’s tool, not as already having shipped one.

That gap matters for how “AI adoption” gets reported. Most industry content treats the two largest operators as interchangeable leaders in AI, because they lead on everything else. On this specific measure they are not tied at all — one shipped a product over a year ago, and the other is still working through a rejected patent application.

FanDuel’s year-plus head start

FanDuel launched AceAI in March 2025, describing it as “the industry’s first generative AI sports betting chat experience.” It lets users ask questions in text or voice to find or build bets, pull team and player statistics, and get responses drawing on FanDuel’s own odds and data. As of mid-2026 reporting, AceAI had processed roughly 158,000 queries and reached about 50% of FanDuel’s user base in a staged rollout — more than a year after its original closed-beta launch. The tool also includes a built-in responsible-gambling layer: it is designed to detect language suggesting a user may be struggling and proactively surface RG resources, escalating concerning cases internally.

Hard Rock Bet is the only other operator in this set with a comparable live feature. Its AI Insights tool launched March 18, 2026 — a full year after AceAI — and takes a narrower approach: rather than an open chat interface, it delivers AI-generated game previews (injury notes, momentum, matchup context) personalized to a user’s favorites and displayed on the app’s “For You” page, initially covering NBA, NHL, and soccer.

“AI-powered” often means something else entirely

Several products that get described in coverage or marketing as “AI-powered” turned out, on closer inspection, not to be generative AI consumer features at all. ESPN Bet’s FanCenter — a personalized hub of props and parlay suggestions tied to a user’s ESPN Fantasy rosters — runs on what ESPN’s own announcement calls “account linking technology,” with no mention of AI or machine learning anywhere in the launch materials. PrizePicks’ widely covered “Feed” feature, despite headlines framing it as an AI play, is a social feature that lets users copy friends’ lineups — again, no generative AI involved. This is the pattern behind a lot of “AI in gambling” content: real personalization and data features get relabeled as AI in secondary coverage, inflating how widespread actual AI adoption looks.

Prediction markets show the same gap from a different angle. Kalshi and Polymarket are consistently covered as the industry’s most AI-native platforms, largely because a wave of third-party tools (trading bots, odds analyzers, browser extensions) has been built on top of their public APIs. Neither platform has shipped its own first-party AI chat or assistant feature — the AI layer that exists sits entirely with outside developers, not the operators themselves.

Offshore sportsbooks: same question, same answer

US sports betting legalization is state-by-state, not federal, and coverage of “AI-visibility leaders” almost always means the licensed side of that split. As of 2026, only 30 states plus DC and Puerto Rico offer legal online/mobile sports betting; 11 states — Alabama, Alaska, California, Georgia, Hawaii, Idaho, Minnesota, Oklahoma, South Carolina, Texas, and Utah — have no legal sports betting option at all, licensed or otherwise. Bettors in those states, and plenty in states with legal options who prefer offshore terms, use offshore sportsbooks instead. Those operators don’t show up in indexes like 5W Research’s, which is scoped to the licensed US market, so we checked them separately, directly against each operator’s own site.

The result: of 8 major offshore sportsbooks used by US bettors — Bovada, BetOnline, MyBookie, BetUS, Xbet, SportsBetting.ag, Everygame, and BUSR — none has a live, named, consumer-facing AI feature. BetUS is the one partial exception worth flagging on its own: it lists an “Artificial Intelligence” category, but that’s a novelty prop-betting market (wagers on AI-related news and outcomes), not an AI tool built into the platform — the offshore-market version of the same mislabeling problem found in Finding 3 above. The full operator-by-operator breakdown is in the results table further down this page.

Casinos and crypto casinos: the same result, for a different reason

We checked six major US-facing licensed casino brands (BetMGM Casino, DraftKings Casino, Caesars Palace Online, FanDuel Casino, Golden Nugget Online Casino, BetRivers) and five widely used crypto-first casinos (Stake, BC.Game, Roobet, Cloudbet, Duelbits) directly against their own sites and recent press coverage, using the same bar as the sportsbook census: a named, publicly launched, consumer-facing AI product. None has one — including FanDuel Casino, despite its sportsbook sibling AceAI being the single strongest live result in Part 1; there’s no indication AceAI, or anything like it, extends into FanDuel’s casino product. Unlike sportsbooks, casino “AI personalization” is rarely a discrete, announced feature at all — it’s typically folded into an unbranded backend recommendation engine (“games you might like”) rather than shipped and marketed the way FanDuel launched AceAI. So the 0-of-11 result here isn’t the same finding as DraftKings’ rejected patent; it’s evidence that the casino side of the industry talks about AI in trend coverage far more than it ships anything specific enough to name and verify.

Fraud and KYC: where AI adoption is already close to universal

The contrast with consumer-facing AI is sharp. According to SEON’s February 2026 fraud-industry survey, 98% of organizations in fraud and AML workflows — a category that includes gambling operators as one of the highest-risk verticals — now deploy AI in some part of that process, and 95% of respondents say they’re confident the AI works as intended. This isn’t a niche investment: 94% of surveyed teams plan to increase fraud-related headcount in 2026 (up from 88% the prior year), and 83% expect their fraud budgets to grow, with 57% of betting and gaming operators specifically reporting that fraud is currently outpacing their revenue growth.

The dollar figures back up the scale of this shift. Future Market Insights sizes the iGaming AI fraud-detection market at roughly $1.9 billion in 2026, growing at a 25% compound annual rate toward a projected $14.7 billion by 2035. On the ground, that shows up as real operating spend: a Tech-Insider vendor survey from June 2026 found mid-sized operators paying $15,000–$40,000 per month for combined AI fraud stacks, with per-verification costs in the $0.80–$3.80 range depending on the depth of the check (Sumsub) and per-API-call pricing around $0.04–$0.10 for lighter checks (SEON). Most of that spend — 57.6% — now goes to end-to-end platforms that bundle multiple fraud functions rather than single-purpose point tools, and 84.8% of deployments are cloud-based.

Put simply: an AI feature that helps a bettor build a parlay is still a novelty two operators have shipped. An AI system that flags a fraudulent account or a stolen card is close to a baseline requirement across the entire regulated industry. Prediction markets fit the same pattern from a different angle: Polymarket and Kalshi have no first-party consumer AI feature of their own — the AI activity around them is entirely third-party tooling (trading bots, odds analyzers) built on their public trading data, not something the operators built or ship.

The AI Adoption Gap Index: quantifying the split

Combining the results of our own census with SEON’s industry-wide back-office adoption figure produces a number nobody else has published: across the 29 consumer-facing gambling operators checked in this census (18 sportsbooks, 11 casinos/crypto casinos), only 2 — 6.9% — have shipped a real consumer-facing AI feature. Industry-wide, 98% of gambling and fintech fraud teams already use AI in some part of their fraud/AML workflow. That’s a 91.1 percentage-point gap between how much AI the industry has actually deployed behind the scenes versus in front of the customer — roughly 14 times more back-office adoption than consumer-facing adoption. We’re calling this the NewBonuses.com AI Adoption Gap Index, and we’ll track it as this data changes: as of August 2026 it sits at 91.1 points.

Player protection: AI is already deciding who gets flagged as at-risk

A smaller but fast-growing category sits between the marketing-heavy consumer side and the near-universal fraud side: AI used to detect problem gambling before a human ever reviews the account. Mindway AI’s GameScanner — described as an “early detection of at-risk and problem gambling” tool that combines AI, neuroscience-based modeling, and expert assessment — is publicly listed as in use by Entain, evoke, BetCity.nl, Betr, ZEAL, JOI Gaming, Bet25, and Omnigame. Independent validation from Gaming Laboratories International found the system matches at least 87% of the at-risk and problem-gambling cases that trained human experts would flag manually — a rare instance in this space of an AI claim backed by third-party testing rather than only the vendor’s own numbers. Mindway AI’s related product, Gamalyze, is used directly by Greece’s Hellenic Gaming Commission for official player self-assessment — a rarer case still of a national regulator, not just an individual operator, adopting this kind of tool.

A separate 2026 industry survey of 350 operators, regulators, and investors found 56% now rate AI’s importance to the industry at 9 or 10 out of 10 (up sharply from prior years), while traditional tools like self-exclusion and deposit caps have shifted from being competitive differentiators to baseline regulatory expectations — with AI-driven behavioral monitoring increasingly filling the gap above that baseline. The same survey also surfaced a more uncomfortable data point: 63% of respondents said responsible-gambling messaging sometimes backfires, and 25% said they felt a stronger urge to gamble immediately after seeing an RG ad — a reminder that this is an area where the technology’s stated purpose and its actual effect on behavior don’t always line up cleanly, and one directly relevant to ongoing scrutiny of how operators collect and use behavioral data in the first place.

How this census was conducted

Part 1 sample — licensed sportsbooks. The 10 sportsbook, daily-fantasy, and prediction-market operators ranked highest for AI citation visibility in 5W Research’s Sports Betting & Gaming AI Visibility Index 2026, which scored 25 operators on how often they were cited across ChatGPT, Claude, Perplexity, and Google AI Overviews for more than 75 consumer-intent betting queries. This index covers the licensed US market only.

Part 1 sample — offshore sportsbooks. 8 offshore sportsbooks commonly identified in industry review coverage as among the most used by American bettors, including those in the 11 states with no legal betting option: Bovada, BetOnline, MyBookie, BetUS, Xbet, SportsBetting.ag, Everygame, and BUSR. No equivalent AI-visibility index exists for this segment, so each operator’s own sportsbook site was checked directly rather than ranked by citation frequency.

Part 2 sample — casinos and crypto casinos. Six major US-facing licensed casino brands (BetMGM Casino, DraftKings Casino, Caesars Palace Online, FanDuel Casino, Golden Nugget Online Casino, BetRivers) and five widely used crypto-first casinos (Stake, BC.Game, Roobet, Cloudbet, Duelbits), checked directly against operator sites and recent press coverage for a named, consumer-facing AI product.

Parts 3–4 sources. Fraud and AML adoption figures are drawn from SEON’s February 2026 industry survey and Future Market Insights’ iGaming AI fraud-detection market sizing; operating-cost figures from a Tech-Insider vendor survey (June 2026) and Sumsub’s published KYC pricing. Player-protection figures are drawn from Mindway AI’s own published operator list and GameScanner validation data (Gaming Laboratories International), and from a 2026 iGaming Trends survey of 350 industry participants reported by PlayUSA.

What counted as a confirmed live consumer AI feature (Parts 1–2). A named, publicly launched, consumer-facing product using generative AI or AI-branded analysis — verified against the operator’s own press release, app-store listing, site content, or direct company statement, not against secondary coverage alone. Back-end uses of AI were intentionally excluded from Parts 1–2 and instead form the subject of Parts 3–4. Third-party AI tools built on an operator’s public data or API were not counted as the operator having an AI feature. Personalization or recommendation features not described as AI by the operator itself were also excluded, even where secondary press coverage characterized them as “AI-powered” — this excluded BetUS’s “Artificial Intelligence” novelty betting market, which is a wagering category, not a product feature.

AI Adoption Gap Index calculation. Consumer-facing adoption (6.9%) is our own census result: 2 confirmed live features out of 29 operators checked (18 sportsbooks + 11 casinos/crypto casinos). Back-office adoption (98%) is SEON’s independently published, industry-wide survey figure for fraud/AML AI use, not a figure we measured ourselves — it covers a broader set of organizations than our 29-operator sample, so the 91.1-point gap should be read as a directional comparison of two real, sourced figures rather than a controlled apples-to-apples measurement of the same population. We think that’s still a meaningful and citable number; we’re also naming the limitation plainly so it’s used correctly.

Limitations. This is a snapshot as of August 2026. Consumer-facing status can change quickly — DraftKings in particular has signaled intent to launch a competing tool. Offshore and crypto casino operators in particular publish little to no product-launch documentation, so absence of a public AI announcement is a reasonable but not absolute indicator that no feature exists. Fraud and player-protection figures are self-reported by vendors and industry surveys rather than independently audited in most cases, with the exception of GameScanner’s GLI-validated detection rate. Status should be re-verified before citing beyond this date.

Full results table — licensed sportsbooks (5W AI Visibility Index)

Full results table — offshore sportsbooks

Player-protection AI — publicly listed operators

Citation & reuse. This census was produced by NewBonuses.com (August 2026), drawing on primary sources including FanDuel’s and Hard Rock Bet’s own product announcements, ESPN Press Room, 5W Research’s Sports Betting & Gaming AI Visibility Index 2026, direct review of each offshore and crypto-casino operator’s site, SEON’s 2026 fraud industry survey, Future Market Insights’ iGaming AI fraud-detection market report, Mindway AI’s published GameScanner client list and GLI validation data, and PlayUSA’s coverage of the 2026 iGaming Trends survey. The findings, tables and charts are free to cite, quote and reference with attribution and a link to this page. Journalists and researchers may contact us through the site for the underlying sourcing.