Personal AI assistants are becoming an integral part of digital life, transforming simple user instructions into complex requests and actions on websites. Whether a user wants to compare products, find support articles, or complete a purchase, they rely on their assistant. For website operators, however, all this interaction often appears as regular automated traffic. Without a clear overview of which assistants are visiting the site and precisely what they are doing, it's difficult to assess whether this traffic genuinely benefits the business or, conversely, creates potential risks. While broadly blocking all automated agents might seem like the safest choice, it can ultimately lead to the loss of valuable customers. The key is to gain more detailed visibility into the activities of these assistants, enabling informed decisions about the most suitable approach for your website.
Thanks to new analytical tools, it's now possible to deeply understand how personal AI assistants access your website on behalf of users. These tools are complemented by granular access policy settings, allowing you to implement these decisions in practice. Together, they empower you to welcome these agents under terms that best suit your business.
Understanding which specific AI assistant is behind a request transforms previously inexplicable automated traffic into data you can work with confidently. Modern security platforms leverage extensive databases of verified bots, and this knowledge is now integrated into detailed overviews of your traffic. As a result, a new category, "Personal Assistants," appears in security analytics alongside verified bots, probable humans, probable automated traffic, and fully automated traffic. Security teams can select this category and monitor only the traffic generated by personal agents over time.
In the most frequent statistics overviews, a new "Personal Assistants" widget further lists which agents visit your website and how many requests each sends. Filtering by a specific agent updates all other widgets to display only the details of its traffic: what paths it requests, what status codes it received, and from which networks it originates. At a glance, you can see if the assistant is logging in, reading product pages, reaching the cart, or encountering errors on pages it cannot use.
For websites protecting user accounts with advanced account abuse prevention tools, operators and fraud prevention teams can see which user each agent represents. A user population overview shows which personal assistants are active across your users and how many accounts utilize them, helping you determine if agent traffic is a meaningful part of your business. Each login and registration event now indicates whether the request was made by a personal assistant and which specific one. Fraud prevention teams can thus filter accounts by events associated with AI assistants or by the number of active AI assistants.
Furthermore, each user account overview displays the assistants acting on behalf of that account. An analyst investigating an unusual login can easily determine if it originated from an assistant the customer has relied on for weeks, or from an assistant that appeared for the first time today and requires closer examination. If an attacker appears to be using an assistant on an account, analysts can immediately mitigate the risk for that account by applying a security rule to its hashed user ID, without blocking the assistant for everyone else legitimately using it.

Brauchen Sie Hilfe mit Ihrer Website oder App?
Melden Sie sich, und ich antworte innerhalb von 24 Stunden mit einem Vorschlag für das weitere Vorgehen.
These overviews provide you with a clearer basis for deciding which personal agents to welcome, restrict, or block.
Access control options have recently been expanded to allow website operators to set specific rules for various types of AI bot behavior, including those used for search, training, and assistant functions. The existing agent policy is now extended with granular control options for supported personal agents. In line with this broader scope, the setting is now called "Access Rules for AI Crawlers and Agents."
In the security settings, you can define a general policy for agents and then choose a different policy for each supported personal assistant: "Allow" (do not block), "Block on ad pages," or "Block." For example, you might set the general agent policy to "Block on ad pages," but for a specific supported personal assistant you wish to welcome, choose "Allow" (do not block). Your existing policies for search and training remain unchanged.
You can also make these choices when setting up a new zone for your website. During the initial setup, we often ask if you monetize pages with ads and subsequently recommend appropriate policies. Of course, you can modify your access settings at any time later in the security settings.
The correct policy depends on your website's specific needs. One publisher might allow a supported personal assistant access across their entire site so readers can easily find their articles, while another might block the same assistant on pages where ads support the content. Both publishers choose the access control that best suits their business.
Agent identity is only part of the story. The analytical tools and policies available today provide website operators with visibility into the activities of personal agents and control over their access. The next goal is to help website operators preserve the context of user relationships when personal agents act as intermediaries. A website may recognize the agent making a request but lacks information it would otherwise have about the user: whether they are an existing customer, what they authorized the agent to do, whether they have a subscription, or what protections should apply.
Achieving this goal will require privacy-sensitive methods to (1) convey user context via the agent, (2) help websites assess reputation, and (3) establish clear accountability when something goes wrong. We will share more information as we collaborate with agent operators and other entities across the ecosystem, so website operators can decide whether to grant agents greater access while continuing to build relationships with their users. The person asking a personal assistant to compare products, find support, or complete a purchase is still a customer trying to reach your business. Review how personal assistants are accessing your website, then visit the security settings and Access Rules for AI Crawlers and Agents to establish the terms that suit your website.
