Plaid launches a tool that tells AI agents from persons at login
Plaid says its tool found 2.3 times more agent sessions than other tools, with no increase in incorrect flags.
Plaid, a company that links finance data to apps, launched Agent Detection for banks and other finance companies. The tool shows if a person or an AI agent drives a session. It also names the AI provider of the agent. Pedro Sanzovo, Head of Fraud at Plaid, says agents are on login, onboarding and account-link pages at this time. This makes it hard for firms to know who acts.
How it works
The tool reads many thousands of signals from devices, behavior, automation patterns and network traffic. It marks each session as person or agent. Plaid says it tuned the model for finance. It can tell an agent that logs in as a finance customer from an agent that shops.
The numbers
In tests on live traffic against other tools, Plaid says its tool found 2.3 times more agent sessions. It says incorrect flags did not increase. The tests are from Plaid. It gave no external test results.
What is not known
Plaid expects that finance firms will have different positions on agents. It plans a method for consumers to give approval for each agent task. It also expands its Trust Index, which scores sessions for risk, to include agent work. Plaid gave no date for these.
Insight: Agent detection vendors race to be the gate
Plaid names the AI provider behind an agent. Incode links an agent to a checked person. This reads as a race to own the gate, because the first gate that firms use can set the rules. Leaders must select a gate before agent traffic grows.
Sources
- Plaid launches Agent Detection for AI agentsThe Paypers
Posted