AI Poker Bots: How They Work, Whether They’re Legal, and How Sites Catch Them

AI can now beat the world’s top poker professionals in heads-up and six-player games. That technology has escaped the lab, as unscrupulous players use a poker bot to play every hand for them at poker sites. Real-time assistance (RTA) tools, which steer every decision a cheat makes, have become an even bigger problem. Reputable poker sites ban both, and their security teams actively hunt down these accounts, close them, and confiscate their funds.

This guide explains how poker bots work, how AI learned to beat the pros, and where poker sites draw the line between cheating and legitimate study. You’ll also learn whether using an AI poker bot is illegal, how sites detect bots, and how to spot one at your table. Our guide to AI in gambling covers how the same technology is reshaping the wider industry.

What Is a Poker Bot?

A poker bot is a program that plays poker hands automatically, without a human making each decision. The bot interprets the cards, betting actions, and other information, then folds, calls, or raises hand after hand. Unlike a human, a bot can play long sessions without fatigue or emotional lapses, often across several tables at once. This sets poker bots apart from study tools, which help you analyze decisions but never make a move on your behalf.

Poker bots range from basic to highly advanced. Older bots follow simple rules written by their creators, such as raising with strong starting hands and folding weak ones, which makes them predictable and easier for experienced players to exploit. More sophisticated AI bots learn their strategy from millions of hands instead of hand-coded rules, and the strongest can hold their own against skilled humans. That’s why poker sites now treat bots as a serious threat to their players rather than a minor nuisance.

Bot farms take the problem a step further. In a bot farm, one person runs a group of bots across different accounts, so the operation can cover far more games than a single player could manage alone. Sometimes several of these bots end up at the same table and share information about their cards, which gives them an unfair advantage over the human players they’re up against. A bot farm therefore cheats in two ways at once: the bots play for their operator, and they team up against everyone else at the table.

How AI Learned to Beat Poker

AI toppled the world chess champion in 1997, but it took another two decades for machines to conquer poker. Humans have kept their edge for so long because poker is a game of hidden information, whereas chess shows everything. Without seeing its opponents’ cards, a poker AI has to make decisions based on incomplete information and learn when to bluff and when to call.

The breakthrough came in January 2017, when Carnegie Mellon University’s Libratus program took on four top poker pros in heads-up no-limit Texas Hold’em at Rivers Casino in Pittsburgh. After 20 days and 120,000 hands, Libratus secured a resounding victory, finishing more than $1.7 million ahead in chips. Between sessions, the program even worked out fixes for the weak spots the pros had started to exploit.

Impressive as it was, Libratus could only beat one opponent at a time. Most real poker games have several players at the table, and every extra opponent multiplies the number of possible situations, so Libratus’ approach couldn’t keep up. Two years later, Carnegie Mellon and Facebook AI cracked the problem with Pluribus, which was built to play six-player games.

Pluribus faced five elite pros at a time, and it won steadily. If the chips had been worth real dollars, it would have won about $1,000 per hour. The pros that it faced included two former World Series of Poker Main Event champions. Even more striking, the researchers trained the Pluribus poker bot on a single server in just eight days, at an estimated cloud cost of under $150.

Both bots learned mainly through self-play rather than by studying human games. In self-play, the AI starts out by playing randomly against copies of itself. After each hand, it reviews its decisions, determines where another option would have paid off more, and adjusts its strategy accordingly. Over an enormous number of hands, that process results in a balanced strategy that bluffs at the right frequency and is very hard to exploit, all without a human ever teaching the AI how to play.

Poker Bots vs RTA vs Solvers: What’s Allowed?

Not all poker software is off-limits, and the deciding factor is usually when you use it. Many serious players now use AI study tools to hone their skills, so it’s important to know where poker sites draw the line. Using AI during a hand is banned at every reputable online poker room, while using it to study between sessions is generally fine. This rule exists to keep the game fair, as you should be playing against other people, not against a program that feeds you answers.

Here’s a snapshot of how the main types of poker software compare:

Of the two banned categories, real-time assistance poker tools are now widely seen as the bigger problem. With RTA poker software, a human still makes every click, so the account plays at a natural pace and behaves like any other player. That makes RTA much harder for sites to detect than a fully automated bot.

Poker sites combat RTA by focusing on the strategy behind the clicks. They flag accounts that match solver play too closely and ask players under suspicion to verify their identity or play on camera. Our section on how poker sites detect bots explains these methods in more detail.

Casino games are a different story, with their own rules on blackjack and roulette software, which our guide to AI casino tools covers in detail.

Are Poker Bots Illegal?

Using a poker bot isn’t usually a crime in its own right, but running one against real players could still count as fraud. In legal terms, fraud generally involves deceiving people for financial gain, and prosecutors could argue that a bot does that by passing itself off as a human player to win other people’s money.

Courts haven’t always agreed that running a bot meets that definition. In one high-profile Swedish case, a group of players convicted of aggravated fraud for running bots on a state-run poker site had their convictions overturned on appeal, after the court found the legal test for fraud hadn’t been met. Some markets leave less room for doubt. For example, it’s an offense to use software to gain an advantage in a licensed game in Nevada. Because the rules vary so much, the safest assumption is that a bot could cause you legal problems.

In practice, the biggest penalties usually come from the poker site rather than the courts. Running a bot or RTA tool breaks the terms you agree to when you sign up. That allows the site to permanently close your account while keeping your balance, including any winnings. As you accepted the terms, you’d have little chance of getting the money back via a legal challenge.

Some major poker sites go a step further by returning the confiscated funds to the players who lost to the cheating account. They sometimes even cover any shortfall out of their own pocket if they can’t claw enough back. A single crackdown can lead to refunds for thousands of players, so if you suspect a bot has cost you money, reporting it is worth the effort. Our section on how to spot a poker bot explains what to look for and how to report it.

How Common Are Bots in Online Poker?

There are no reliable figures on how many bots play online poker. Anyone who gives you a precise number is simply guessing. Bots are built to blend in, and poker sites don’t share complete data on the accounts they catch. However, ban reports give us a sense of the problem, as several major online poker rooms now publish snapshots of the accounts they’ve closed for using bots or RTA.

One major site closed 161 accounts in a single year and refunded more than $360,000 to over 4,000 affected players. Another banned 98 bot accounts and refunded around $154,000 to players. These reports only cover the cheats who got caught, so the true number of online poker bots is likely higher.

Bots can be a bigger concern at crypto poker rooms. Many of these sites let you sign up and play with little or no identity verification. This appeals to privacy-minded players, but it also makes the sites easier for bot farms to exploit. Without ID checks, one person can open several accounts and run a bot on each, and banned players can more easily come back under a new name.

That doesn’t mean every no-KYC poker room is riddled with bots, but it does make a site’s approach to security more important. Before you deposit, look for sites that publish their security and fair-play policies, explain how they handle bots and RTA, and, ideally, share their own ban reports. Our guide to anonymous crypto casinos covers how no-KYC sites work in more detail.

How Poker Sites Detect Bots

Poker sites use a mix of behavioral analysis and security checks to catch bots and players using RTA. Unusual playing patterns will often trigger an investigation, and further checks help confirm whether the account is cheating:

• Timing analysis: Sites track how long each account takes to make decisions. Humans normally spend longer mulling over tough decisions and act quickly on easy ones, while bots often act at unnaturally consistent speeds, and RTA users can show odd pauses in spots that should be routine.
• Session tracking: Online poker rooms monitor how many tables an account plays, for how long, and at what times. Plenty of regulars play several tables at once, so volume alone isn’t proof of anything, but an account that plays heavy volume around the clock with barely a break, day after day, looks far more like a bot than a person.
• Solver comparison: The sites compare players’ decisions against the solver’s strategy over thousands of hands. Strong players often match it in simple spots, but an account that keeps playing near-perfectly in complex situations raises a major red flag, and this is one of the main ways sites catch RTA. However, they weigh this alongside other signals, as some elite players can also play near-perfectly.
• Performance tracking: Poker sites measure each account against its own history, so a win rate that jumps overnight, or play that suddenly becomes much stronger, stands out.
• Player reports: Many investigations start when other players notice something odd and report it.
• Software and device checks: Poker sites often scan your device for known automated or assistive programs. They can also link accounts run by the same person, which helps them uncover bot farms.
• Verification checks: Sites can ask suspicious accounts to complete identity checks, pass in-game challenges, or play a session on camera. Bots struggle with these checks, and so do RTA users, as their play often drops off sharply once they’re being watched.

How to Spot a Poker Bot at Your Table

You won’t always be able to tell the difference between a bot and a human, but there are a few red flags to watch out for:

• Always online: The same account seems to be playing whenever you log in, at any hour, often across several games at once. Regulars can put in long sessions at an online poker room, but no human can realistically play around the clock for weeks on end.
• Identical timing: The account takes the same amount of time to act, regardless of whether it’s folding a weak hand or facing an all-in on the river. Most people slow down when a decision gets tough, whereas bots tend to play like metronomes.
• Unvarying bet sizes: The account bets the same amounts in the same spots every time, regardless of the board or the players involved. This is typical of simpler, rule-based bots.
• No chat or reaction: The account never responds to chat, never sits out, and never times out, even after a painful loss.

These signs don’t offer conclusive proof that you’re facing a bot. After all, plenty of quiet, disciplined players display some of these habits. RTA users are even harder to spot, as a real person is making every click.

If you suspect cheating, don’t confront the player at the table. Instead, send the account name and the relevant hand numbers to the site’s security team. The team can then analyze the data and mount a full investigation. If the site bans the account, you could be in line for a refund, so keep a note of your report and follow up if you don’t hear back.

Using AI to Improve at Poker the Legitimate Way

You can use the same technology that powers poker bots to become a much stronger player without breaking any rules. Just make sure you only use it to hone your skills between sessions, and always switch it off before you play.

Solvers are the most sophisticated study tools. A GTO solver works out a balanced strategy for a specific situation, so you can load a spot you struggled with, see how often the solver bets, checks, or bluffs, and pinpoint where your own play went wrong. Trainer apps take a more hands-on approach, testing your decisions in common spots with drills and quizzes, and then showing you how much each mistake costs.

AI chatbots can help too. You can paste in a hand and ask for a street-by-street analysis of your decisions, or ask a chatbot to explain concepts like pot odds, ranges, or bet sizing in plain language. Just bear in mind that chatbots can get poker math wrong, so treat their answers as a starting point and check anything important against a solver. If you’d rather practice than study, plenty of free apps let you play against bots without risking any money, which is a useful way to test new ideas before you try them out in cash games and tournaments.

The rule of thumb for AI poker tools is simple: study with AI, but play without it. These tools are generally fine to use between sessions, but the moment one starts guiding your decisions during a hand, it crosses into RTA territory, and that can get your account closed and your balance confiscated.

The Bottom Line on AI Poker Bots

With AI now strong enough to beat elite pros, bots and RTA tools are a significant threat at online poker rooms. However, poker sites are using increasingly sophisticated methods to catch and ban these accounts, giving you more protection than you might think. Stick to reputable poker sites that publish robust security and fair-play policies, as that’s where these safeguards are most reliable. Some sites even refund the players who lost out to bots, so it’s always worth filing a report if you suspect cheating.

You can also sharpen your play by harnessing the same technology that makes bots so powerful. Just follow this key rule: study with AI, play without it. For a wider look at how AI is changing everything from sports betting to casino games, check out our guide to AI in gambling.

FAQ