TL;DR
Math.tanh and other mathematical functions in browsers can reveal the underlying operating system due to differences in their implementations. By analyzing the output of these functions, researchers found that OS-specific libraries produce varying results, which can be exploited for fingerprinting.
✦ Why It Matters
Developers should consider implementing additional checks to mitigate OS fingerprinting risks in web applications.
Key Takeaways
Full Summary
Fingerprinting techniques typically focus on more obvious signals like canvas and WebGL, but recent findings reveal that mathematical functions like Math.tanh can also leak OS information. Different operating systems utilize distinct mathematical libraries, leading to discrepancies in function outputs.
For instance, macOS uses Apple's libsystem_m, while Linux relies on glibc, resulting in about 25% disagreement on inputs. Windows, through the Universal C Runtime, shows variations as well.
Testing on Chrome 150 demonstrated that while most inputs yield consistent results, certain values like tanh(0.8) can distinctly identify the OS. This subtlety in mathematical computation presents a new avenue for enhancing browser fingerprinting techniques, raising concerns about user privacy.
Related