Indeed. It probably wouldn’t go over very well if FF surreptitiously rolled out similar ad features through their own experiments infrastructure.

I’m still learning about Google’s federated learning and don’t understand how ad platforms will identify so called cohorts in other browsers

FLoC: for interest-based audiences. The API generates clusters of similar people, known as "cohorts". Data is generated locally on the user's browser, not by a third party. The browser shares the generated cohort data, ...

It appears there are many working proposals, ideally we could voice an opinion of which is worst for the sake of the unconfigged web.

https://web.dev/digging-into-the-privacy-sandbox/

Additional Info:

• FLoC from the horses mouth: Federated Learning of Cohorts: https://github.com/WICG/floc
• Turtledove: Another google proposal: https://github.com/WICG/turtledove

[FLoC] addresses category, ads targeting based on someone's general interests. For personalized advertising ... please check out the TURTLEDOVE proposal.

• FLEDGE: https://github.com/WICG/turtledove/blob/main/FLEDGE.md

with some bizarre features such as on-device bidding and bring your own untrusted server:

On-device bidding by buyers (DSPs or advertisers), based on interest-group metadata and on data loaded from a trusted server at the time of the on-device auction — with a temporary and untrusted "Bring Your Own Server" model, until a trusted-server framework is settled and in place.

On-device ad selection by the seller (an SSP or publisher), based on bids and metadata entered into the auction by the buyers.

• Privacy Sandbox Explainer: https://github.com/bslassey/privacy-budget

• Draft Spec: https://wicg.github.io/floc/

wrt to Firefox, it is interesting to a see a Mozilla rep commenting in the issues https://github.com/WICG/floc/issues/4

For example, based on Firefox telemetry data the IP addresses of a percentage of our users changes regularly over time. With FLoC those requests that would be presented to the network with different IP addresses ... (by Ehsan Akhgari ehsan, Mozilla https://github.com/ehsan)

and different browser vendors in the draft spec:

The string representation of the interest cohort version is implementation-defined. It’s recommended that the browser vendor name is part of the version (e.g. “chrome.2.1”, “v21/mozilla”), so that when exposed to the Web, there won’t be naming collisions across browser vendors. As an exception, if two browsers choose to deliberately use the same cohort assignment algorithm, they should pick some other way to give it an unambiguous name and avoid collisions. (https://wicg.github.io/floc/#interest-cohort-section)