Rep. Josh Gottheimer (Democrat, New Jersey-5) and other members of the United States House of Representatives have filed a bill that would require prediction markets to use scans of app and website users’ faces to ensure that they meet the platforms’ age minimums. The “Facial Recognition to Protect Children Act” perfectly fits the definition of a misnomer, but the legislation has the support of Kalshi.
SEON Vice President of Risk & Compliance Nauman Abuzar says that facial scans have some use in terms of age estimation, but that usefulness could be most limited in the exact area of emphasis for prediction market age requirements.
Prediction markets can account for the limitations of the technology in their compliance systems. However, the technology is not the game-changer that will eliminate the risk of underage trading and give loving homes to all the puppies in all the shelters in the whole world that Gottheimer is selling.
H.R. 9706 shows poor comprehension of “facial recognition”
As Abuzar stated, “age estimation and identity verification are different problems.” Facial recognition is concerned with identity, not age estimation.
Identity verification is about using the scan to enhance trust that a person actually is who they are presenting themselves as. The system evaluates whether the face being presented matches the stored data about that individual face.
Age estimation systems, on the other hand, measure whether a face fits the model’s parameters for the target age group. Regardless, the text of Gottheimer’s H.R. 9706 and his office’s press release about the bill consistently uses the term “facial recognition.”
There could be some element of wanting to use language common to the popular lexicon to identify the technology, but with technical requirements such as those the bill proposes, accuracy is important. The misuse of terms aside, the bill’s tenets could be useful in a limited fashion.
What H.R. 9706 means for prediction markets
H.R. 9706 dictates that “a wagering operator or a prediction market platform operator may not permit a user to access a wagering or prediction market platform under the control of the wagering operator or the prediction market platform operator, or accept a wager, in the case of a wagering operator, or place an order, in the case of a prediction market operator, from such user, that the wagering operator or the prediction market platform operator has not verified, using commercially available facial” scan “technology, has attained more than 18 years of age.”
There are also sections of the bill concerned with privacy.
“A wagering operator or a prediction market platform operator may not collect, process, or transfer the covered data of a user beyond what is reasonably necessary, proportionate, and limited to the purposes for verifying the age of the user.”
“A wagering operator or a prediction market platform operator shall delete any covered data collected with respect to a user that the wagering operator or the prediction market platform operator determines is not necessary for compliance with the requirements of this subsection.”
Abuzar believes that it’s possible for prediction market exchange operators to meet those requirements.
Modern facial scans could satisfy H.R. 9706’s privacy demands
Firms that have designed facial scan technology have accounted for the privacy concerns that Gottheimer’s bill addresses.
“There are solutions today that are designed to perform age estimation or facial verification without retaining the original biometric data once the process has been completed,” Abuzar explained. If the objective is simply determining whether someone appears to be above a certain age, retaining identifiable biometric information isn’t always necessary after that assessment has been completed.”
As to the reliability of that estimation, that depends heavily on context, with context including the target age group and the reliance of the scan data that the entity deploys. Discussions of accuracy and reliability wade into the limitations of facial scan technology.
Facial scans are useful for estimation, not proof of age
Prediction markets currently use age verification systems that may use facial scans on a voluntary basis as well as the verification of government-issued documentation. Facial scans may have the most benefit in terms of augmenting those measures rather than replacing them.
“Age estimation technology has improved considerably over the past few years, but it’s important to remember what it’s designed to do,” Abuzar commented. “It’s estimating age, not proving it. As a result, it works best as part of a broader trust framework rather than as a standalone decision-making tool. Combining age estimation with other contextual signals gives organisations greater confidence while reducing reliance on any single control.”
Incorporating age estimation scans into a broader framework is also important for prediction markets because of the level of accuracy that regulations demand. A person on their 18th birthday becomes eligible to trade on prediction markets whereas the day before that birthday they were ineligible.
Facial scan technology is not yet able to reliably make such minute estimations.
“The closer someone is to the minimum age threshold, the more difficult age estimation becomes because people develop differently and there can be significant variation between individuals of similar ages” Abuzar added. “That’s one reason organisations often use additional verification when someone is estimated to be close to the required age, rather than relying solely on the age estimate itself.”
Because of the consequences for non-compliance that H.R. 9706 recommends, exchange operators need to establish that they have performed required scans. That spurs new consideration of privacy concerns.
Can exchanges establish compliance while still respecting privacy?
In the instance that systems fail and a person who is underage is able to access prediction market exchanges, a primary consideration for exchange operators will be to establish their good-faith efforts to utilize required measures to prevent that activity. At the same time, those liability concerns could infringe on privacy interests, but a balance of the two is possible.
“Organisations do have legitimate compliance and audit obligations, but retaining raw biometric data isn’t necessarily the only way to meet those obligations,” Abuzar elaborated. “Many organisations instead retain evidence that verification took place, along with the outcome, timestamps and audit records, without storing the underlying facial images themselves. Modern verification platforms can generate audit records that show when a verification took place, what checks were performed and the outcome of those checks. That allows organisations to demonstrate that appropriate verification processes were followed without necessarily retaining sensitive biometric images.”
Additionally, Gottheimer’s bill wouldn’t require prediction markets to perform a facial scan upon each use of the apps or websites. Scans of that frequency wouldn’t be necessary to improve accuracy or reliability, either.
“A risk-based approach is generally a more balanced model,” Abuzar said. “Rather than treating every interaction the same, organisations can apply additional verification when the level of risk increases. That might include larger transactions, attempts to withdraw funds, changes to account details or activity that doesn’t match a customer’s normal behaviour.”
Given that H.R. 9706 adds new requirements for platforms like Kalshi, Kalshi’s support for the bill might seem out of character. However, a look at the legislation makes it painfully clear why Kalshi CEO Tarek Mansour was present at Gottheimer’s press conference announcing the legislation.
H.R. 9706 contains gift to Gottheimer campaign donors
Tucked into H.R. 9706 is a section that could make it easier for the Commodity Futures Trading Commission (CFTC) to approve gaming-related event contracts like those based on sports and defend itself from litigation over those contracts. Gottheimer’s bill amends part of the Commodity Exchange Act (CEA) to give federal regulators increased flexibility in approving contracts and clarify Congressional intent that the federal government is the sole regulator of event contract trading.
That language in the bill comes after Gottheimer received $18,000 in campaign donations from Coinbase and Kalshi investor Morgan Stanley added another $10,000. Gottheimer is up for reelection in November but his seat representing New Jersey’s fifth Congressional district for a sixth consecutive term has been rated as “safe” by Polling Source.
Gottheimer’s bill could lead to some reduced underage use of prediction market platforms but age estimation with facial scans is not the superhero busting through the wall to save the day that Gottheimer is making it out to be in order to pay back his campaign donors. Accurate and reliable systems would use it as one tool in a more complex array of checks.
“The conversation shouldn’t really be about facial recognition,” Abuzar summarized. “It should be about how organisations build confidence that the right person is using the right account at the right time. Facial verification is one tool that can contribute to that, but the strongest approach combines risk-based verification, good governance and privacy-conscious design to build trust without creating unnecessary friction.”
More on Kalshi
