Normal anti-fingerprinting measures aren’t effective against companies with lots of resources. Does setting the security level to “safest” actually thwart more advanced tracking? The ideal case is that it would make a user look like everyone else who browses on that OS and browser in safest mode.
Safest
- This level only allows website features required for static sites and basic services. These changes affect images, media, and scripts.
- Javascript is disabled by default on all sites.
- Some fonts, icons, math symbols, and images are disabled.
- Audio and Video (HTML5 media) are click-to-play.
Based on the Tor Project’s wording it seems security modes are primarily made to benefit your security and are less about anti-fingerprinting, though I’d assume it helps with anti-fingerprinting since websites are far less capable to do things in general without JavaScript, certain fonts, etc.
Answered here:
This is interesting, but it still leaves my original question unanswered, which is whether this will help against advanced fingerprinting. Like if I go to a website that uses couldflare.com and cloudflareinsights.com, and I’m in safest mode, will disabling all JavaScript help against that? How difficult will I be to distinguish from other people on the same OS using the same browser in safest mode if I go to a website that uses cloudflare and cloudflareinsights?
I ask because I’ve read that normal anti-fingerprinting isn’t enough to protect me in many cases: Brave is NOT fingerprint resistant enough - #4 by Valynor
Like I said in the other post, you do gain some increased protection against fingerprinting, since a lot of stuff is not able to be run at all, but if this is enough for the potentially smaller buckets you fall into 
Well for one the user agent is spoofed in safest mode. JavaScript is disabled and cannot execute code to deanonymise you. Your exit IP will also be different from anyone else. Canvasing is not accurate. Fingerprinting/deanonymising a Tor user is not an easy task.
Honestly I would say visiting a clearnet website on the Tor browser, in safest mode is very safe. I don’t have any reason to believe they could effectively fingerprint you. I’ve never heard of this tactic being used to deanonymise people.
What an adversary would realistically do is try to escape the browser sandbox and execute an IP ping from your host. This is how Dutch police caught many Hansa admins (people now use Tails/Whonix now to mitigate this attack). Or if you’re a person of significant interest multiple agencies from different countries could collaborate and attempt to perform a traffic correlation attack, which would take probably years and a lot of resources.
In conclusion fingerprinting a user in the Tor browser, especially in safest mode, has not proven to be an effective method of tracking or deanonymising users. It’s just not seen in the the wild. At least as far as I know. If anyone knows otherwise please let me know.
You can use services like creep.js to see if your fingerprint ID carries across multiple sessions. Simply due to a different exit IP alone, it shouldn’t correlate and identity you as the same user.
No?
Your entire post relies on a number of broad assumptions and treats fingerprinting as if it happens in a vacuum.
Oh thanks you’re right. I was wrong there. User agent is no longer spoofed now as of about a year ago, as some changes were made.
I still stand by the broader point though. Could you elaborate @any1 on what you mean about treating fingerprinting like it’s in a vacuum?
First of all, the security levels were not meant to increase fingerprinting protection, they are intended for security and the fingerprinting effects are just a side effect.
You are also assuming that running the Safest setting would definitely protect you, without considering the real world, where people have various hardware configurations and different language needs. (What good is it to get the English version of a site if the user can’t understand English? It would turn their browser into a glorified image viewer.) Just because you disable JavaScript, WebGL, etc., doesn’t mean you aren’t unique enough given the information that is still available.
First of all, you cannot stop fingerprinting, you can only try to prevent re‑identification across different sessions. Treat a session as an identity that is linked to the sites you interact with during that specific session. Tor Browser’s goal is not to prevent these identities from being created, but rather to prevent them from being linked to previous identities that exist.
Let’s look at examples of how using Standard and Safest could realistically change this. (The bucket numbers have no meaning other than being an identifier for these examples).
Tor Browser set to Standard - default value, little breakage, so likely the largest user base.
Session 1 (Bucket 54):
- wikipedia
- privacyguides
Session 2 (Bucket 54):
- abcnews
Someone else’s Session 1 (Bucket 54):
- cnn
- amazon
You have now created two sessions that are linked to bucket 54. If the bucket is large enough, it should not be possible to link these two sessions to a specific user because many other users also produce sessions in that bucket.
Tor Browser set to Safest - non-default value, major breakage, so likely a much smaller user base. Because almost everything is unusable, you must add exceptions for some sites, which then relies on the same fingerprinting protections Tor Browser provides on Standard anyway.
Session 1 (Bucket 32):
- wikipedia
- privacyguides
Session 2 (Bucket 27):
- wikipedia
- reddit (does not work on
Safest, so you have to add an exception) - privacyguides
Someone else’s Session 1 (Bucket 32):
- wikipedia
- codeberg
You have now created two sessions that land in different buckets, with Session 2 (Bucket 27) being in a subset of the Safest bucket. Safest plus an exception is different from simply visiting the site on Standard or Safest with no exception. Unless you run a Safest session with no exceptions, you can potentially shoot yourself in the foot, something that would not happen on Standard.
TL;DR: Treat the security slider as a security slider, not as a tool to increase fingerprinting protection. Most people should leave it alone.
Disclaimer: I am just a grifter, so do not take this as the absolute truth 
The comment I linked said that big companies, at least, and possibly others, will be more than capable of re-identifying people using fingerprinting-resistant browsers for the foreseeable future, so that in reality people visiting Reddit and Amazon won’t actually all go into Bucket 54. There’s also Cloudflare, which I mentioned because it’s a sizeable company used by a lot of small websites, and avoiding it isn’t as simple as avoiding big-company websites.
Do you agree with this?
I’ve also seen it said that if I can stream video on a website, I’m not actually anonymous. Is this true as well?
My comment was focused specifically on Tor Browser, which the comment you linked to specifically recommends to use.
Not sure what you exactly mean with this?
I read that comment as saying, “Use Tor, but know that it won’t provide anonymity against big companies with lots of resources to dedicate to tracking you.” There’s a comment under it which does imply that Tor/Mullvad browser is good for anonymity, which is confusing.
I think the idea is that something about streaming video to my computer makes it possible for a website to uniquely identify it regardless of what browser I use. I wish I could find where I read this.
Setting it to “safest” should definitely also be treated as a tool to increase fingerprinting protection. The information a website can gather in safest from a tor browser session is extremely reduced compared to the default setting. I am certain that this outweighs the lesser user adoption and that enough tor browser users use safest to still have a significant net benefit.
If you keep it per session and don’t start adding exceptions for sites, then maybe depending on your setup. You can ask the Tor Browser devs directly and they will confirm that it was never meant to be a slider for increasing fingerprinting protection, since standard should be strong enough to protect you.