While process automation technology providers identified financial services as a ripe target in the early days of RPA, most systems using AI agents have not focused on the space so far. According to a new report from Human Security, however, agentic AI traffic targeting financial services websites more than doubled from April to May, signaling growing experimentation with AI-driven workflows across financial services platforms of all kinds.
Human, a New York City-based security and antifraud company, said traffic to financial services sites increased 124 percent in May, although the category still represented less than one percent of overall observed agentic activity.
The report suggests financial services could become a larger target market for AI vendors building autonomous agents for enterprise workflows. Companies including OpenAI and Anthropic are developing agent-based systems aimed at financial tasks and decision support.
“The considerable increase by percentage—though small in actual volume—continues an ongoing trend observed by Human, indicating that AI agents are becoming increasingly active across banking, investments and insurance,” the company said.
Human also said the increase in agentic activity is creating new operational and security challenges. Because AI agents can behave similarly to automated malicious traffic, some detection systems may struggle to distinguish legitimate activity from abuse.
Overall agentic traffic declined 4.3 percent month over month in May, according to the report, though the blocking rate rose from 8.2 percent in April to nearly 9 percent. Human said this indicates a gradual increase in policy controls governing AI-driven browsing activity.
The company identified Comet Browser and Atlas as the largest sources of agentic traffic in May, though both declined in absolute volume. Meanwhile, the Claude Chrome Extension and Genspark gained share during the month.
Media, e-commerce and travel sites accounted for nearly 98 percent of observed agentic activity.