WPT Global has banned 575 accounts flagged as tournament bots, according to the analyst who reported them — the largest single poker bot purge any online room has publicly acknowledged.
The report came from Vojtech Cervinka, who runs the X account MTT Database Review and has spent years using tournament databases to identify accounts whose play patterns are statistically impossible for humans. His post was blunt: three months ago he reported 575 accounts to WPT Global as tournament poker bots. They banned every single one.
Why 575 Is a Big Number
To understand the scale, it helps to compare it to what the industry has previously admitted to.
WPT Global's first-ever public cheating report, covering January to May 2025, disclosed dozens of banned accounts and $166,885.07 seized and redistributed to affected players. That was considered an unusually transparent disclosure at the time — the room published 50 internal account IDs, which almost no operator does.
The 2024 ACR Poker Venom episode, still the most-discussed bot incident in recent memory, involved an account called "Withia" that reached second place overall after Day 2 of a $12.5 million guaranteed tournament before being locked and blinded off. Three more suspect accounts — "Verucasalt5," "MReGoR" and "acquiton" — were locked in the same sweep, and roughly 50 previously reported MTT bots were banned alongside them. Cervinka was the source on that one too.
At the time, Cervinka wrote that "MTT streets look pretty clean today," while adding that the work was not finished and that the industry needed "more transparent refunds" and "better security processes." That was fifty accounts. This is 575.
How the Detection Actually Works
There is a persistent misconception that bot detection is primarily a technical exercise — that operators scan for automation software, unusual mouse movements, or virtual machine signatures. Those methods exist and they catch the crude operations.
The sophisticated detection is statistical. Tournament databases record every hand a screen name plays across every tracked event. A human player's decision timing varies with the difficulty of the spot; a bot's does not, or varies in patterns that repeat. A human's strategy drifts over months as they study; a bot's is frozen at the moment it was deployed. Groups of accounts sharing a strategy profile that no human population would independently converge on are, in practice, one operator running a farm.
Cervinka's method is closer to forensic accounting than to antivirus scanning. He builds profiles from public tournament data, finds clusters, and hands operators lists. The operators then confirm with their own server-side data — IP addresses, deposit methods, device fingerprints, session overlap — and act.
This division of labour is worth noticing. The most effective anti-bot work in online poker over the past several years has been done by an independent analyst with a spreadsheet, not by the security departments of the rooms themselves. Matt Berkey, speaking about the ACR case, described Cervinka as a regular contributor in a high-stakes Discord with direct lines to operator representatives, and noted that Cervinka had warned WPN about an incoming bot attack more than a month before the Venom began.
What This Means for Players
Bots concentrate where the structure suits them, not where the stakes are highest. Bot farms target low-to-mid buy-in multi-table tournaments with large fields and long registration windows, because those events let a single operation enter hundreds of accounts across a schedule with minimal per-entry scrutiny. If you play $5 to $50 online MTTs in volume, you are in the affected population. High-stakes cash games are, counterintuitively, less exposed — the fields are small, the regulars know each other, and anomalies get spotted within days.
Room selection is a security decision, not just a rakeback decision. The rooms worth playing are the ones that publish enforcement data, seize funds from banned accounts, and redistribute them. WPT Global's disclosure practices are not perfect but they are considerably better than the silence most operators maintain. Our safe poker sites guide weights security posture and payout history above bonus size, which is the right ordering.
Refunds are the unresolved problem. Banning accounts is the easy half. Working out who lost money to those accounts, across how many tournaments, and returning it accurately is genuinely hard — and most operators do it slowly, opaquely, or not at all. When you evaluate a room's response to a bot incident, the question is not "did they ban the accounts" but "did affected players get paid." Cervinka has been consistent on this point for years.
Watch for the tells you can actually see. You will not out-detect a statistical model at the table. But repeated identical bet sizings from an opponent across many hands, instant actions in genuinely difficult spots, and screen names with similar construction patterns seated across multiple tables are worth noting and reporting. Room support teams take structured reports seriously; vague accusations they ignore.
The Structural Argument
There is a broader point here that the poker industry has been slow to confront. Bots are not a technology problem that will be solved with better detection. They are an economics problem.
A bot farm is profitable when the expected value per account per tournament, multiplied by the number of accounts, exceeds the cost of running them and the risk-adjusted cost of getting caught. Detection raises the last term. It does not eliminate the equation. As solver-derived strategies become cheaper to obtain and cloud compute gets less expensive, the cost side falls every year.
The counterweights that actually work are structural: aggressive know-your-customer requirements at signup, limits on multi-accounting from shared infrastructure, and — the one operators dislike — reducing the number of tables a single account can play, which cuts the throughput a farm can achieve. Rooms have been reluctant to impose these because they also inconvenience high-volume legitimate regulars, who are among the most valuable customers.
The 575-account ban is therefore worth reading as a data point about scale rather than as a victory announcement. If one analyst working from public data could identify 575 bots on one site, the total population across the industry is considerably larger than 575.
Where This Leaves Recreational Players
Realistically, the risk to a casual player entering a handful of tournaments a month is small in expectation and large in perception. Bots do not target individuals; they harvest small edges across enormous volume. The player genuinely harmed by bot activity is the serious grinder playing forty tournaments a week at buy-ins where farms operate.
That said, the correct response is the same for everyone: play at rooms with published enforcement records, understand the rakeback and bonus terms you are actually getting, and read reviews that cover security rather than just promotions. Our reviews of Americas Cardroom, BetOnline and Black Chip Poker each address network security policy alongside the games and traffic.
And if you play primarily tournaments, it is worth knowing which series have historically drawn bot attention — the very large guarantee events with long late registration are the ones farms plan around.
Cervinka's post ended without triumphalism. The work continues.
Sources: PokerOrg, Pokerfuse, Poker Industry PRO