You cannot delete a browser fingerprint the way you delete a cookie, because it is built from details your browser reports every time it loads a page. You can, however, make it far less useful. The goal is not to be invisible but to look like many other people, or to change often enough that you cannot be followed.
Two strategies: blend in or randomise
Blending in means reporting the same values as a large group, so that you are one of thousands. Tor Browser does this by standardising window size, fonts and many other values. Randomising means adding small random changes so that your fingerprint differs on each site or session. Brave takes this approach for canvas and audio. Both work; mixing them badly does not.
Steps that help
- Pick a browser with built-in protection. Tor Browser for the strongest result, Brave for convenience, or Firefox with strict tracking protection.
- Turn on the stricter mode. In Firefox, set
privacy.resistFingerprintingto true in the advanced configuration page. It rounds the window size and time zone, and it can break sites that depend on them. - Keep the window at its default size. A maximised or oddly sized window is itself a signal; Tor shows a warning for this reason.
- Use few extensions. Each extension can be detected, and a rare combination is distinctive.
- Block third-party scripts. Most fingerprinting is done by third-party trackers, so a good content blocker removes many of them before they run.
- Update your browser. Vendors keep reducing what is exposed, for example by shortening the user agent.
What does not help
- A VPN alone. It changes your IP address, not your fonts, screen or graphics card.
- Private or incognito mode. It clears cookies and history, but reports the same details.
- Random user agent switchers. A user agent that disagrees with the rest of your browser stands out, and can be more identifying than the real one.
- Clearing cookies. Irrelevant to a fingerprint, which does not use them.
Test before and after
Run the browser fingerprint test in your current browser, note the bits figure and whether the canvas is flagged as randomised, then repeat after each change. A lower figure and a randomised canvas show the protection is working. Try the test in a second browser to see how much difference the choice of browser makes.
The trade-offs
Strong protection costs convenience. Resist-fingerprinting modes can show the wrong time zone, force light mode and trigger more CAPTCHAs. Some banks and fraud-detection systems treat a very generic or noisy browser as suspicious, so keep a separate normal profile for those sites. Decide where the protection matters most, such as casual browsing and research, and accept some friction there.