Experts from Consult Hyperion, Prove, NIST, iProov identify biometric binding, liveness among key tools
AI agents are dominating conversations about technology and commerce, and accordingly many sessions at FIDO’s Authenticate 2026 focused on agentic authentication and authorization.
A panel discussion on the third day of the event moderated by Consult Hyperion Principal Consultant Steve Smith looked at the issue from the perspective of liability.
K&L Gates Partner Judie Rinearson, who specializes in payments, points out that right now legal protections against agents making fraudulent online purchases come from credit card regulations.
But as Prove GM for New Market Initiatives Frances Zelazny points out, credit cards aren’t strongly bound to an identity.
There is also a whole body of law about delegation, though it was formulated with real estate agents and other human representatives in mind, rather than AI agents. In those cases, evidence about the transaction is compared to the applicable laws and policies.
But identity evidence is typically presented in the form of civil or national IDs. But ID documents have known fraud challenges, and national identity databases occasionally suffer breaches.
Hence, the reliance on authentication methods that are known not to work, like passwords and OTPs. Zelazny says binding a biometric to the account is the answer.
Rinearson is happy to see the industry working on protocols related to agentic evidence, but regulation will be needed to support trust in the long term, she says.
