The new offering is a way to control artificial intelligence-driven automation by connecting it with “real, verified human identity” within Sumsub’s know your agent (KYA) framework, the company said in a Thursday (Jan. 29) news release.
“As AI agents and browser-based automation grow in popularity, businesses struggle to tell legitimate activity from fraud,” the release said. “Most platforms treat automation as inherently suspicious and block it by default. In light of this, Sumsub’s AI Agent Verification enables businesses to separate lawful, human-driven automation from malicious agent attacks by linking all activity to a verified human identity. This creates a clear line of accountability, allowing legitimate automation to operate while preventing illicit activity.”
The tool treats automation as a “manageable risk” and not an immediate red flag, according to the release. The system first determines when activity is automated, gauges its risk level, and only applies additional checks when needed.
In high-risk scenarios, AI Agent Verification can ask for a targeted liveness test to make sure that a real human is present and authorized. In doing so, it blocks deepfakes from being used in place of real users and makes sure that every action is tied to the person responsible, the release said.
“Rather than attempting to blindly trust AI agents themselves, our solution focuses on verifying the humans behind them,” Sumsub co-founder and Chief Technology Officer Vyacheslav Zholudev said in the release.
Advertisement: Scroll to Continue
KYA technology is important because agents behave exactly like the patterns fraud systems are designed to mistrust, PYMNTS reported this month.
“They operate continuously, retry efficiently and optimize relentlessly for completion,” the report said. “Without identity and mandate, those behaviors are indistinguishable from bot attacks or account takeover attempts.”
Fraud systems are built to be conservative. When they lack context, controls compensate by rejecting more transactions. Agent-initiated commerce exacerbates that bias because automation looks unusual by default. Without KYA, issuers notice volume without any explanation, and the rational response is to clamp down on controls.
“What makes KYA viable now is timing,” the report said. “The identity infrastructure required to support it has matured.”
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
