Stock image, for illustration only. Not related to the actual events.
ESPN’s broadcast of the 2026 World Series of Poker Main Event featured a new artificial intelligence tool designed to detect player tells and predict hand strength, but professional players interviewed by WIRED are skeptical the system has enough data to be effective.
The AI tells detection tool appeared during the first days of the Main Event broadcast in early July, displaying live metrics on player movements alongside a hand-strength model breaking down the probability of different hand types. Created by Luke Geel, an AI engineer for the US Air Force, the system analyzes eye movements, blink rate, posture, chip handling, and hand fidgeting to predict whether a player holds a strong hand, a drawing hand, or is bluffing.
The tool sparked immediate debate in the poker community about whether it represents a novel broadcast enhancement or an intrusive attempt to digitize a fundamentally human skill.
Limited Data Raises Questions
The system was trained exclusively on camera footage from the 2026 Main Event, which drew over 9,000 entries. Only three tables were recorded throughout the tournament, meaning the vast majority of players never appeared on camera long enough for the AI to build a meaningful dataset on their behavior.
“The streams are varied enough that you don’t get the same players too frequently,” said Michael Gagliano, a 17-year poker professional who reached this year’s final table and is playing for the $10 million top prize. Gagliano told WIRED he reviewed every second of ESPN’s live streams during the two-and-a-half-week break after the final table was set in mid-July, searching for tells on his remaining opponents, but found the limited screen time constrained what he could learn.
Any AI analyzing the same footage would face identical limitations, he noted, even for players who spent considerable time on camera during deep tournament runs.
Pros Doubt AI Can Match Human Tell-Reading
Professional players emphasized that tell detection requires understanding context and intention, not just cataloging visual patterns. Shaun Deeb, a two-time WSOP Player of the Year winner who finished 15th in the 2026 Main Event, described the range of tells available to observant players as far broader than cameras can capture.
“There are leg tells, checking tells, verbal tells, breathing tells, pulse tells,” Deeb said. “There’s an insane amount of tells available, and most of those can’t be picked up by a camera.”
Even if the tool accurately identified when a player projected confidence or weakness, that alone wouldn’t decode their actual hand. Gagliano pointed out that the same hand strength can produce different reactions depending on the player, the stakes, and the situation. A casual player might feel extremely confident holding two pair, while the same hand in a high-stakes Main Event could still provoke nervousness despite being strong for the situation.
“Maybe my body language is referencing the situation rather than the hand strength,” Gagliano explained.
Mixed Results and Uncertain Future
Geel acknowledged the tool’s limitations in an email to WIRED, saying he’s conducted blind tests on other poker competitions with mixed results and that a larger sample of hands would improve performance.
Deeb was blunt in his assessment of the tool’s broadcast value. “I think they randomly found something to try to make it like another sport, and I just think it was swing-and-a-miss,” he said.
While the AI feature appeared during portions of the July broadcast, a representative from Omaha Productions, which is licensed by ESPN for WSOP and other sports coverage, confirmed in a text message that the tool would not be used for the final table. The representative declined to explain the decision.
As AI technology advances, even skeptics concede that tell-detection tools could evolve and eventually be applied for competitive advantage. High-roller tournaments with six-figure buy-ins feature a small pool of recognizable professionals who play each other repeatedly in events that are often broadcast, potentially creating hundreds or thousands of hours of footage on individual players. Studying such footage is already common practice among pros preparing for major events.
Deeb said he routinely hires live-tells specialists to observe opponents during deep Main Event runs, and a close friend watched the streams on his behalf during his 2026 run. For now, he’s confident human analysis still holds the edge over any algorithm.
Key questions answered
Why do professional players think the AI tell-detection tool doesn’t have enough data to work effectively?
The system was trained only on footage from the 2026 Main Event, where just three tables were recorded out of over 9,000 entries. Most players never appeared on camera long enough for the AI to build a meaningful dataset on their behavior.
What kinds of tells do pros say the AI tool can’t detect through cameras?
According to Shaun Deeb, cameras can’t pick up leg tells, checking tells, verbal tells, breathing tells, and pulse tells, which are all available to observant players at the table.
Will ESPN use the AI tell-detection tool during the World Series of Poker final table?
No. A representative from Omaha Productions confirmed the tool would not be used for the final table but declined to explain the decision.
Who created the AI tell-detection tool shown during the broadcast?
Luke Geel, an AI engineer for the US Air Force, created the system that analyzes eye movements, blink rate, posture, chip handling, and hand fidgeting.