NEW YORK — The artificial intelligence summary is the first thing anyone sees after conducting an online search. The content that follows those summaries are increasingly AI-driven. That creates complex trust scenarios where artificial intelligence agents act autonomously on behalf of humans while learning and adapting to anticipate user preferences without direct oversight.
“Over 50% of all traffic on the internet is no longer human. It’s now machine based,” Stu Soloman, CEO of HUMAN Security, told Beet.TV Editorial Director Lisa Granatstein at the IAB Connected Commerce Summit. “A subset of AI traffic is agentic traffic. I would define agentic traffic as traffic that has the authority and autonomy to act on behalf of a human, and then further starts to continuously learn and adapt and anticipate what that human would have wanted.”
This evolution moves beyond binary good-versus-bad bot scenarios toward nuanced trust evaluation as agents increasingly conduct business without humans directly in decision loops.a
Agent-to-agent commerce emerges
Artificial intelligence agents representing consumers now communicate directly with merchant agents, placing humans at arm’s length on both transaction sides while raising accountability questions when problems arise.
“Increasingly, agents representing consumers are talking to agents representing merchants with no human directly in the loop and a human at arms length on both sides of that relationship,” Soloman said.
Identity verification becomes complex when true human identity may not exist behind agent interactions, replaced by machine identities that require new trust frameworks for transaction accountability.
Brand protection beyond discovery
Companies must evolve from protecting intellectual property and brand presentation toward ensuring transaction integrity as agent interactions progress from discovery through conversion.
“Brands are increasingly now focused on not just protecting their intellectual property or the discovery and presentation of their property and their brand equities, but now also understanding that a transaction at the end of it was the one they wanted and expected,” Soloman said.
This progression requires brands to build agent interaction capabilities that maintain brand affinity while ensuring trusted conversion outcomes.
Siloed controls create vulnerability
Success in AI environments demands integrated customer journey management rather than discrete control points across marketing, trust and safety, fraud prevention, and security functions that treat singular customer interactions as separate processes.
“All the controls that we put in place are very siloed in nature,” Soloman said. “A customer in an AI world does not. It’s a singular interaction from the time that you’re interfacing with digital content, to the time that you get attracted to that brand, then you click on it, to the time that you authenticate and ultimately that you transact.”
Traditional discrete control expectations fail to address unified customer experiences that span from digital content engagement through authentication and transaction completion.
“This is a value stream or a customer journey that’s singular in focus,” Soloman said, “but we treat it with discrete control points that have to change.”
