Washington’s seriously eyeing facial recognition technology as a frontline defence against underage gambling. A new bipartisan bill would require age verification checks every single time someone logs in or places a bet on prediction markets and sportsbooks.

Congressman Josh Gottheimer introduced the Facial Recognition to Protect Children Act back in July, and he’s got backing from industry players including Kalshi CEO Tarek Mansour. The bill tackles a real problem: right now, verification only happens when you create an account. Which means a teenager could easily borrow a parent’s or older sibling’s device and start wagering with virtually no friction.

The Scale of the Problem

The numbers are pretty sobering. More than a third of boys aged 11 to 17 have gambled in the past year, according to research cited in the bill. State regulators are catching cases all the time. Iowa’s Division of Criminal Investigation has fielded dozens of underage betting reports. Tennessee sportsbooks shut down over 400 accounts connected to minors in 2024. Those are just the incidents discovered within the regulated framework.

Americans wager around $160 billion on sports annually, generating $16 billion in revenue. That’s a substantial financial incentive for platforms to look the other way. A quick technological fix could save operators from both regulatory headaches and genuine reputational damage.

The Technical Reality Check

This is where things get messy. Facial recognition algorithms can estimate age, but they’re hardly infallible. Poor lighting, camera quality, distinctive facial features. All of that can confuse the system. People near the legal threshold will create edge cases constantly.

Plus, security researchers have shown that circumventing facial checks isn’t exactly rocket science; video game avatars have fooled the technology before.

Privacy Concerns Loom

The bill’s supporters insist that biometric data won’t be stored long term. But processing those images still requires moving them somewhere, even temporarily. Cybersecurity experts point out that automated systems alone aren’t enough. When decisions carry real consequences, human review becomes necessary. And frankly, no software is perfect.

For operators, this proposal represents a genuine attempt to close a compliance loophole that regulators are increasingly focused on. Whether facial recognition actually proves reliable enough in practice? That’s the fundamental question.