LexisNexis Brings 'Know Your Agent' AI Detection to Government

LexisNexis Risk Solutions recorded 750% growth in agentic traffic across the global digital transactions it analyzed during the first half of 2026 — and on November 10, 2026, it responded by bringing Know Your Agent (KYA) intelligence to government agencies through its ThreatMetrix and BehavioSec products.

The numbers explain the timing. AI agents — software that acts on a person's behalf, filling forms, logging into accounts, initiating payments — are moving from novelty to routine participant in online services. Government systems, which handle benefits, enrollments and public funds, now face a question their identity controls were never designed to answer: is the thing on the other end of this session a human, a conventional bot, or an AI agent acting for a citizen?

What does the new capability actually do?

Inside ThreatMetrix and BehavioSec, KYA pulls together four kinds of intelligence — digital identity, device, behavioral and network signals — to spot recognized AI-agent signatures and other indicators of agent-driven activity. The mechanism is straightforward: agents leave patterns, and those patterns can be matched against known signatures.

The signal doesn't operate alone. KYA combines it with transaction intelligence inside the risk-assessment and decisioning workflows agencies already run. That means an agency sees agentic activity in the context of the wider interaction, not as an isolated flag.

There's a reason this matters now. Many personal agents do not currently declare themselves as agents, according to the company's announcement. An agency can't simply ask "are you a bot?" and trust the answer, because the software on the other side often has no obligation — or mechanism — to say so. Detection has to happen through behavior and infrastructure signals instead.

The takeaway: KYA turns agent presence into a data point inside existing fraud and identity workflows, rather than a separate system agencies must bolt on.

Is this about blocking AI agents?

No — and the company is explicit about that. The announcement frames the agent signal as one consideration in a broader assessment, "rather than a universal instruction to block automation."

Agencies keep control of the response. They can decide how to handle each interaction according to their own policies, their risk tolerance, and the sensitivity of the service or transaction involved. The signal complements existing identity verification, authentication, fraud-prevention and cybersecurity controls.

The supported interactions listed in the announcement show the range at stake:

• Benefits and program access
• Digital-service enrollment
• Authentication and account access
• Payments and disbursements
• Individual self-service

The same signal could therefore inform two very different decisions: a routine account login and a public-funds disbursement. The response would vary, because the risk varies. A permissive policy for self-service chat doesn't have to mean a permissive policy for moving money.

The takeaway: detection is a dial agencies can set per transaction, not a switch that bans automation outright.

Why isn't detection enough?

Here's the harder problem, and the announcement spends real time on it. Government identity and authentication controls were built to establish one thing: that the person using an online service is who they claim to be.

An AI agent breaks that model by introducing another participant. Knowing an agent is present tells an agency what is acting — but not who it represents, and whether its actions are authorized. Those are separate questions, and for sensitive transactions they're the ones that carry the consequences.

Haywood Talcove, CEO of LexisNexis Risk Solutions Government, put the stakes plainly: "The real measure is whether an agency can protect the citizen, the transaction and the public funds behind it."

That framing matters because it moves the goalpost. The practical test for agencies is delegated authority — whether an agent may enroll someone in a program, access an account, or initiate a payment on a person's behalf. Identifying the agent supplies context; the decision that protects citizens and funds is the authorization call.

The takeaway: spotting the agent is step one; proving it has permission to act is the decision that actually protects public money.

Where did KYA come from?

The government announcement follows a broader launch. On November 5, 2026 — five days earlier — LexisNexis Risk Solutions announced Know Your Agent for its digital identity and fraud-prevention platform, adding agentic insights for organizations generally, not just the public sector.

That earlier release included a notable detail: many customers were already using KYA intelligence in risk-management decisions before the government push. The November 10 announcement extends that existing capability into government availability, agency-specific workflows and public-service applications.

The sequence suggests demand preceded the product's government debut. Commercial customers were already folding agent signals into their risk decisions, and the public-sector rollout answers the same need where the stakes are public funds rather than private revenue.

The takeaway: KYA is a five-day-old commercial product that already has users, now pointed at government.

Will the approach hold up as agents evolve?

LexisNexis describes KYA as a protocol-agnostic foundation — meaning it isn't tied to one agent framework or technical standard. The company says it can evolve alongside agents, standards and government use cases, including environments where autonomous systems interact with government systems and with one another.

That last scenario is worth watching. Agent-to-agent interaction — a citizen's agent negotiating directly with an agency's agent — would compound the identification problem the November 10 announcement is built to address. A detection layer anchored in behavior and network signals, rather than a fixed protocol, is the company's hedge against that shift.

For agencies, the near-term homework is policy, not procurement. KYA supplies the signal; each agency still has to decide, under its own risk tolerance, when an agent may act on a citizen's behalf and when it may not. The vendors can flag the agent. Only the agency can grant it authority.

The takeaway: the tool is built to adapt, but the authorization rules — the part that guards public funds — remain each agency's job to write.