Two weeks after Meta launched its Muse personal AI agent in September, traffic matching the agent had appeared at more than half the enterprise customers Cequence Security monitors. Most would not have known they were hosting it.
Cequence tracked the agent across financial services, retail, travel and software customers. At the median customer, Muse traffic grew nearly sixfold in roughly two weeks.
The agent does not announce itself. It runs a real Chrome browser instance hosted by Meta, routes through a consumer VPN, and presents no headers that would identify it as artificial. Cequence identified it through behavioural analysis, including a browser build version that moved from 0% to more than 90% of Muse traffic within days — a pattern consistent with a centrally managed fleet, not organic user updates.
At financial institutions, Muse completed multi-factor authentication on behalf of users. A successful MFA login no longer confirms a person is on the other end. Muse also searched, compared options, filled carts and moved through checkout, with a small share of sessions completing purchases — removing the completed transaction as a reliable signal for distinguishing bots from customers.
In late September, one travel and hospitality customer began blocking nearly one in five Muse requests, targeting only those that looked abnormal rather than banning the agent outright. Agent Trust, a new capability Cequence released this week, addresses exactly that need.
Agent Trust pairs behavioural detection with identity verification. It maintains a live inventory of all agents in a customer's traffic, including those that do not self-identify, and allows security teams to stop specific requests rather than the agent session as a whole.
"AI agents like Meta's Muse are bots that act on behalf of real customers. Many businesses can't see them, and those that can are tempted to block them," said Hari Nair, VP of product management at Cequence. "Agent Trust lets businesses block the action, not the agent. The agent keeps working for its customer, and the one request that crosses a line gets stopped, slowed or challenged."
The product is available now within Cequence Application and API Protection. The analysis covers traffic from 1 September to 24 September 2026 across Cequence's customer base.
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
