GA4 AI Referral Regex Builder

How to use

1. Tick the assistants to include. Add any other source host you see in your reports.
2. Choose whether to match subdomains such as www.
3. Paste a few Session source values from your Traffic acquisition report into the test box.
4. Copy the regex into a GA4 report filter or into a custom channel group condition.

What it checks and how it works

The regex is a list of exact host names joined with a vertical bar and with the dots escaped, for example chatgpt\.com|gemini\.google\.com. GA4 matches the whole value by default, so no anchors or wildcards are needed. The list covers chatgpt.com, gemini.google.com, perplexity.ai, claude.ai and the Copilot hosts copilot.microsoft.com and copilot.cloud.microsoft. Microsoft's documentation names copilot.cloud.microsoft as the primary Copilot URL; copilot.microsoft.com is on Zaher's list and was not confirmed on a Microsoft page, so check it in your own Traffic acquisition report. The tool does not match Bing, which is a search engine.

The list is Zaher's and it will go out of date. Google's own default AI Assistant channel uses a list that its help page does not publish, so use this regex to build a view you control and add the source values you actually see. Google's pages state a 250-character limit for regexes in report filters; the tool shows the length of yours. The test runs your pattern in the browser's JavaScript engine, while GA4 uses RE2, so confirm in GA4.