ServicePipe fingerprint script uses names that trip model safeguards
Original title: Why Is This Fingerprinting Script Creating Bioweapons?
After crawling 10,000 sites, the author found hidden Unicode in a live script that Astra refused to analyze and GLM-5.3 did not.
A researcher crawled 10,000 Tranco sites with Playwright, hooked canvas fingerprinting calls, and published 1,033 renderings at fingerprint.garden. One canvas came from a ServicePipe script at servicepipe.tech whose function names refer to ricin and aspirin lethality. The author reads those names, plus a string of invisible Unicode characters, as anti-LLM obfuscation meant to block deobfuscation. Astra refused the analysis on biological-safety grounds; GLM-5.3 complied with the same prompt.
Unconfirmed: The anti-LLM purpose is the author’s interpretation; ServicePipe has not confirmed it in the text.
- The crawl filtered canvas toDataURL calls and properties such as navigator.webdriver.
- Known canvases included FingerprintJS and Castle among other vendors.
- The ServicePipe file path given is /static/checkjs/06b667/4CIrx8U3Da.js.
- Daybreak Blue on Astra also blocked the deobfuscation request.
- The author was still inspecting what the invisible Unicode string represented when the excerpt ends.
- Vendors
- ServicePipe
- Products
- fingerprint.garden
From the original article
7 min read at nullpt.rs
I've always had a fascination with reverse-engineering browser fingerprinting scripts. Most of my time has been spent looking at the usual suspects (popular fingerprinting libraries like FingerprintJS and more sophisticated scripts used by anti-bots like Cloudflare Turnstile, Akamai, and others). At some point, I wanted to know what else was out there.
By hazebooth · nullpt.rs
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.