As much as I hate whatever google's doing, this article has some issues:

>For operations that need Play Integrity attestation specifically, a compliant Android device costs approximately $30 at current market prices

This assumes the logic on google's side is something like `if(attestationResult == "success") allow()`, but it's not hard to imagine the device type being factored into some sort of fraud score. For instance, expensive devices might have a lower fraud score than cheaper devices, to deter buying a bunch of cheap devices. They might also analyze the device mix for a given site, so if thousands of Chinese phones suddenly start signing up for Anne's Muffin Shop, those will get a higher fraud score.

>Firefox for Android does not appear in Google’s stated browser support list for Fraud Defense.

The browser only needs to show a QR code, so if you're on firefox mobile they'll either open a deeplink to google play services on the phone itself, or show a qr code.

>One human solving a single challenge pays a negligible cost. A bot farm running concurrent sessions faces exponential compute costs with each additional attempt - and AI agents, which consume GPU cycles to operate, face identical penalties regardless of how sophisticated their reasoning is.

PoW for bot protection basically never caught on because javascript performance is poor, and human time is worth more than a computer's time. An attacker doesn't care if some server has to wait 10s to solve a PoW challenge, but a human would. An 8-core server costs 10 cents per hour on hetzner. Even if you assume everyone has a 8-core desktop-class CPU at their disposal (ie. no mobile devices), a 6 minute challenge would cost an attacker a penny. On the other hand how much do you think the average person values 6 minutes of their time?