Google Patents a System That Tracks Users Without Ever Seeing Who They Are
Google has filed a patent for a system that can generate consistent user tracking codes without any single party, including Google itself, ever seeing the raw personal data underneath. It's a privacy-by-design approach to a problem the ad industry has been wrestling with for years.
How Google's surrogate ID system hides your real identity
You're signing up for a service, and the company wants to match your account to an ad profile, but nobody is supposed to actually read your email address or phone number. That's exactly the kind of problem this patent tries to solve.
Google's system takes your encrypted personal data and, inside a locked-down computing environment (think of it as a sealed room that even Google engineers can't peek inside), turns it into a surrogate ID: a consistent stand-in code that can be used for matching and targeting without your real identity ever being exposed in readable form. The encryption keys needed to do this are split across several independent "key coordinators," so no single party holds enough pieces to decode your data on their own.
The result is a system where advertisers and publishers can still do audience matching, but the underlying personal data stays encrypted throughout the entire process. Whether that's enough to satisfy regulators or privacy advocates is a separate conversation, but the technical design does add a meaningful layer of protection compared to passing around plain-text email lists.
… decrypting, by the surrogate ID generator, the encrypted user data using the first key to obtain user data in cleartext; generating, by the surrogate ID generator, a surrogate ID using the user data and the second key; …
Translation: The system temporarily unmasks the hidden user data inside a secure zone to build a replacement ID.
How the TEE assembles keys without any one party holding both
The patent describes a surrogate identifier (ID) system with three main moving parts.
First, the system receives encrypted user data from a requesting entity, such as an advertiser or publisher. The data arrives already encrypted, so the system never sees the raw personal information.
Second, a surrogate ID generator runs inside a trusted execution environment (TEE), which is a hardware-level isolated zone on a server chip where even the operator of the machine cannot inspect what's running inside. This generator reaches out to multiple independent key coordinators, each of which holds only a fragment of two separate cryptographic keys. The generator assembles the full keys only inside the sealed TEE:
- Key 1 decrypts the incoming encrypted user data into readable form, but only inside the sealed environment.
- Key 2 is then applied to that readable data to produce the surrogate ID, a consistent code that represents the user without being the user's actual identity.
Finally, the surrogate ID (not the decrypted personal data) is sent back to whoever requested it. The split-key design means no single coordinator, and not Google itself, can reconstruct both keys independently, which limits the blast radius if any one party is compromised.
A surrogate ID generator running in a first trusted execution environment (TEE) of the surrogate ID system obtains, from multiple key coordinators, respective parts of a first key and respective parts of a second key.
Translation: A special secure processor gathers puzzle pieces for two keys from separate managers to unlock the process.
What this means for ad targeting after third-party cookies
For regular users, this is mostly invisible infrastructure. But it sits at the center of a multi-billion-dollar industry question: how do you run targeted advertising after third-party cookies disappear, without creating a surveillance free-for-all? A surrogate ID system like this lets ad platforms do audience matching and attribution without passing around raw email addresses or phone numbers in the clear.
Google's steady investment in privacy-preserving ad tech reflects pressure from regulators in Europe and the US, and from Apple's App Tracking Transparency rules, which have already reshaped how mobile ads work. If this system or something like it ships in a real product, it could become the plumbing that powers ad matching for years, which is precisely why the design choices here carry weight beyond a single patent.
That makes this Google's 43rd filing we've tracked since May in our on-device AI privacy watchlist, adding to earlier applications like writing emergency messages on-device and AI insights pushed to home hubs.
Splitting a key across multiple independent parties means no single company can decrypt your data alone, which is a real structural guarantee. The cost is fragility: if any one of those parties goes offline or pulls out, the whole system stalls. For identity infrastructure handling sensitive data at scale, that's probably the right tradeoff, but it's a bet on sustained cooperation among parties who may have competing interests.
The deeper limit is that the system protects your personal information on the way in, then hands companies a consistent code that follows you around afterward. That code can still be used to build a detailed picture of your behavior over time.
The engineering here is careful and the privacy gain is real, but partial. Locking the front door while leaving the back door a policy question is a meaningful choice, and buyers of this technology should know which problem it actually solves.
There are more where this came from
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.
The drawings
3 drawing sheets from US 2026/0270058 A1 · click any drawing to enlarge
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →