Government agencies may use LexisNexis Risk Solutions' new Know Your Agent (KYA) capabilities to assess AI-driven activity in benefits, tax and licensing services, adding a signal to existing digital-risk reviews.
LexisNexis Risk Solutions said the capabilities are available through its ThreatMetrix and BehavioSec products. Agencies can consider agent activity alongside information about identity, devices, behavior and transactions. Each agency would decide how to handle an interaction under its own policies and risk assessment.
The company cited benefits, tax systems, licensing portals and other digital public services as potential settings. Other examples included enrollment, authentication, account access, payments and disbursements. The announcement named no government customers or deployments and gave no pricing or detection-accuracy rate.
LexisNexis says some personal AI agents do not identify themselves as agents, making it harder for organizations to tell what is generating activity on their services. The company's stated aim is to help agencies protect citizens, public services and funds while allowing legitimate AI use.
Identifying agent activity does not show who an agent represents or establish that it has authority to take a particular action, LexisNexis said in a broader product announcement on Nov. 5. Agencies would need to consider that context when deciding how to respond.
The government announcement followed the company's Nov. 5 introduction of Know Your Agent capabilities for organizations generally. LexisNexis said some customers were already using the capabilities in risk-management decisions, but did not name them.
LexisNexis said its analysis of global digital transactions found agentic traffic rose 750% in the first half of 2026. The Nov. 5 release described growth as more than 750%. Neither announcement gave an absolute count for agentic traffic or explained how the figure was measured.
Haywood Talcove, CEO of LexisNexis Risk Solutions Government, said, “For government, success cannot be reduced to a detection percentage.”
