Two-fetch diff

The two-fetch diff

The two-fetch diff requests the same URL in the same second twice: once as a rendering browser and once as each AI agent. It shows the status codes, bytes and visible text side by side, so you can see the exact sentences ChatGPT, Claude and Perplexity never receive.

User agents per probe: 10
Weighted signals: 51
Fetches compared: 2; Same URL, same second

Run a free diagnosis

How it works

Our probe fetches your page as a browser with JavaScript and as each agent user agent without it. Text is decoded and compared at raw byte offsets using the same container on both sides.

What it measures

Status per agent
Main text ratio
First content offset
Sentence survival
Verbatim claims that survive, get cut by truncation, or only exist client-side

Example: genspark.ai, 2 Oct 2026

Browser: HTTP 200, 206,972 bytes. ChatGPT-User, Claude-User and Perplexity-User: HTTP 403. The diff makes that gap visible in one view.

FAQ

Questions people ask

Can I run a two-fetch diff for free?

Yes. Every free diagnosis at bullai.tech/audit includes a two-fetch diff of your homepage.

Keep reading

LLM LabRead Two-fetch diffRead Prompt BenchRead Answer OwnershipRead Ship and verifyRead PricingRead

Updated 2026-10-13. Sources: Bull AI Labs probe runs, saved per diagnosis · Bull AI Labs saved probe, genspark.ai, 2 Oct 2026