Somewhere today, someone may call a corporate help desk, sound convincing, and get access they should not have. That is the risk Caryn Seidman Becker puts at the center of the AI debate: not whether machines can act, but whether companies can prove the right human asked them to.
Becker, founder, chair and CEO of CLEAR, says documents and credentials were never built to prove identity. Driver’s licenses were created to show you can drive, and Social Security numbers were created to track wages. Over time, those tools became stand-ins for online verification, even as data breaches and generative AI made impersonation easier.
Her argument is not that AI should slow down. It is that AI agents are already moving from writing content to taking action. According to Becker, AI agents can buy products, move money and open sensitive systems for users. That creates what she calls a major opportunity for efficiency, speed and innovation, but it also raises the stakes for identity.
The examples she cites are familiar to any executive who has watched security become a budget line instead of a strategic asset. Google, Visa and Mastercard are building systems that would let AI agents make payments and access credentials like credit cards. Becker says that if a fraudster impersonates an executive on an account-recovery call, takes over the account and registers an agent, fraudulent transactions can follow even when the security controls technically work.
The problem, in her view, is that the control is only as strong as the first identity decision. Today that decision often rests on a password, a security question and a phone call. None of those methods prove it is the right person, she writes. That is why proof of identity, not a separate compliance exercise, becomes the foundation of agent security.
CLEAR says it is a certified Qualified Anti-Terrorism Technology under the Department of Homeland Security. Becker says the company has spent more than 16 years learning that true and trusted identity can strengthen security and deliver faster travel, safer experiences, less friction and better days.
Her broader point is aimed at the way AI is being framed in public debate. The choice is often described as speed versus safety. Becker says that is the wrong tradeoff. In her view, trusted identity lets companies keep both. Slowing AI down is not the answer, she writes, and neither is letting agents run on weak credentials.
That is the market lesson hiding inside the hype. New technology does not erase old incentives; it makes weak verification more expensive. If identity is not tied to action, then fraud gets a faster tool and businesses inherit the damage.
Capital rewards clear rules. When companies build AI on stronger proof of identity, they reduce friction, protect systems worth hundreds of millions of dollars, and make automation easier to trust. When they do not, they are building a skyscraper on sand.
