trust-token-api

The intention of the API is for that to not be possible. > The privacy of this API relies on that fact: the issuer is unable to correlate its issuances on one site with redemptions on another site. If the issuer gives out N tokens each to M users, and later receives up to N*M requests for redemption on various sites, the issuer can't correlate those redemption requests to any user identity (unless M = 1). It learns only aggregate information about which sites users visit. https://github.com/WICG/trust-token-api?tab=readme-ov-file#p...

it's unfortunate that the sour response to this impressive privacy work (https://github.com/WICG/turtledove/blob/main/FLEDGE_k_anonym...) will likely lead to people turning it off, and buried inside is the switch for Private State Tokens (https://github.com/WICG/trust-token-api/blob/main/README.md)

According to the comments in the history of that user on Github, it is someone who claims to be an engineer from Facebook in an earlier post: https://github.com/WICG/trust-token-api/issues/28#issue-6447...

> somewhere between 0 and 64 bits from the implementations I've heard about ... 64 bits is way too much information I think maybe you're confusing this some other API? Here are the trust token docs: https://github.com/WICG/trust-token-api#overview

Much better article: https://www.zdnet.com/article/google-launches-chrome-extensi... Trust Token Github page: https://github.com/WICG/trust-token-api I'm glad to see it uses Privacy Pass as a basic mechanism.

I'm really confused. I'm reading the spec[0], which seems to explicitly state that it's not possible (modulo some possible vulnerabilities with mitigations that they later list) to track a user across the web using these tokens. Yet many of the comments I read here are talking about this being a way for websites to track users? This spec document also doesn't really list use cases for this. Why would I want this? The linked article seems to suggest it's so Google can reduce ad fraud, which of course I don't give a damn about, and don't want my user agent doing anything that helps advertisers, ever. Hopefully Firefox does not implement this, and hopefully if they don't, websites don't somehow penalize browsers that don't support it. [0] https://github.com/WICG/trust-token-api#overview

From https://github.com/WICG/trust-token-api > Token Exhaustion Attack > Issuers issue many tokens at once, so users have a large supply of tokens. > When the issuer detects a site is attacking its trust token supply, it can fail redemption (before the token is revealed) based on the referring origin, and prevent browsers from spending tokens there. I think there’s an interesting trade-off between attack surface and specificity of the token here. A token meaning “This user has only accessed 6 sites in this network” is meaningful only if there’s a single token, as the next access can cause the token to be inconsistent with reality (It states a falsity now). At the same time, it states the issuer can block redemption based on the referrer. I think that’ll be interesting, as an allow list format will lead to only “big players” being able t

https://github.com/WICG/trust-token-api#overview For what I understand, a provider who trust a user, stores in her browser a set of kind of certificates, that basically declare "I "provider NNN", states that this user has been authenticated by me". Another party will ask for one trusted token, verifies that it is really signed by provider NNN, and can trust that each time this token is resent, it is the same user. The privacy is respected on the user side has the third party has no idea who is this user. The trusted tokens could even have a short life. Third parties therefore should have great trust in the trust token provider. It is as if Google authorized other parties to use its own cookies. Indeed this means that other parties trust "provider NNN". I guess that Google wants to become a more central identity provider than today, where