Cloudflare adds per-agent analytics and access controls for personal AI assistants

Cloudflare has launched new analytics and access controls aimed at personal AI agents — assistants that act on behalf of a person by comparing products, digging up support articles, or completing purchases on a site. According to the Cloudflare blog, the goal is to give site owners a clearer basis for deciding which agents to admit, restrict, or block, rather than treating all automated traffic the same way.

Spotting personal agents in traffic analytics

Cloudflare says it is bringing intelligence from its verified bot directory into traffic details. In Security Analytics, the Bot Analysis chart now includes a "Personal Assistants" category alongside Verified Bot, Likely Human, Likely Automated, and Automated, and security teams can select it to view only personal agent traffic over time.

A new Personal Assistants widget in Top statistics lists which agents are visiting a site and how many requests each one sends. Filtering on a specific agent updates every other widget to reflect only that agent's traffic: the paths it requests, the status codes it receives, and the networks it originates from. According to Cloudflare, this makes it possible to see at a glance whether an agent is logging in, reading product pages, reaching checkout, or running into errors on pages it cannot use.

Agent visibility per user account

For sites protected by Account Abuse Protection, which Cloudflare notes is currently in early access, site owners and fraud teams can see which user each agent represents. A population overview page shows which personal agents are active across a site's users and how many accounts use them, and every login and sign-up event now indicates whether a personal agent made the request and which one. Fraud teams can filter accounts by AI assistant events or by the number of AI assistants in use.

Each user's account overview also lists the agents acting on that account. An analyst investigating an unusual login can check whether it came from an agent the customer has relied on for weeks or one that appeared for the first time that day. If an attacker appears to be using an agent on an account, the analyst can mitigate that account directly with a WAF rule tied to its Hashed User ID, without blocking the agent for everyone else who depends on it.

Granular access policies

Cloudflare introduced AI bot policies over the summer, letting site owners make separate access control choices for AI bots in Search, Training, and Agent categories. The Agents policy is now being expanded with granular options for supported personal agents, and the setting has been renamed "AI Crawler and Agent access policies" to reflect the broader scope.

In Security Settings, site owners set an overall Agents policy and can then choose a different policy for each supported personal agent: Allow (do not block), Block on pages with ads, or Block. Cloudflare gives the example of setting the overall Agents policy to Block on pages with ads while allowing a specific supported agent across the site. Existing Search and Training policies remain unchanged.

The choices also appear when a new zone is added to Cloudflare. During onboarding, the company asks whether a site monetizes pages with ads and tailors its recommended policies to the answer, and the controls can be revisited later in Security Settings.

Why it matters

Traffic from a personal agent looks like automation to a site but represents a real customer trying to comparison-shop, find help, or buy something. Cloudflare's argument is that blocking agents outright can mean losing customers, while permitting them blindly makes it hard to tell helpful activity from risk. Per-agent analytics combined with per-agent policies let publishers and retailers make that trade-off explicitly — one publisher might let a supported agent read articles anywhere, while another blocks the same agent only on ad-supported pages.

Cloudflare also frames the launch as a first step. Agent identity, it says, is only part of the story: a site may recognize the agent making a request but still not know whether the person behind it is an existing customer, what they authorized the agent to do, whether they have a subscription, or which protections should apply. The company's next goal is preserving that user context when agents act as intermediaries, which it says will require privacy-aware ways to carry user context through an agent, assess reputation, and establish accountability when something goes wrong. Cloudflare says it will share more as it works with agent operators and others across the ecosystem.