• 0 Posts
  • 3 Comments
Joined 2 years ago
cake
Cake day: June 7th, 2024

help-circle

  • I’ll caveat this by saying IANAL. But the way I read Bill 26-051 is that it’s looking to implement “user age attestation” not “device or application” (WEI). Two separate things.

    Age Attestation requires the OS (or really, the cloud service that implements account-level authorization) and come up with an “age signal.” It prohibits using third-party non-public data, and puts the burden on the OS for managing the Go/No Go process. No PII leaves the device.

    The alternative is dystopian, poorly managed KYC/AML over-reaches. Under the guise of anti-fraud/anti-gambling, these will reach deep into our communal shorts. They could well soon require individual biometric verification (iris scans, face contour maps, fingerprints, etc). No, thanks.

    WEI is a separate story. It’s trying to cut down on malicious apps and maybe stop individual sites doing browser fingerprinting. It can only work on systems with single-points of app installation (without side-loading) and devices already locked down with hardware TPMs. So far, that only covers iOS. All the other systems (Linux, Mac, Windows, and Android) let you install your own system-level code without having to go through the One Official appstore. And with WASM, the browser makes it all moot.

    Personally, I think WEI is a total waste of time. Trying to squeeze the toothpaste back into the tube. But it’s solving a different problem than age verification.

    Not to say the Colorado bill is perfect. There is a truck-sized app vs. website loophole in it, so kids can still access social media sites from the browser vs their phones. But the OS can offer an API that browsers can vend to websites without every site rolling their own crappy system. It also doesn’t account for a clever kid figuring out how to create a separate adult-appearing user account. Because of course, they will.

    Saying it’s parental responsibility is unrealistic. I’ve helped folks set up Screentime, router-level filters, and even Circle (in-home ARP spoofing box, and mobile VPN + fine-grain URL filtering). There are ways around all of it. Besides, the kids can still get exposed to utter bilge via school-approved sites like Zoom, YouTube, or Google Drive. Let’s not even bother with messaging apps or in-game chat. This is all assuming parents have the time or knowledge to set things up and manage the filters.

    We’re not trying to be over-controlling, stop the kids from dancing too close at the prom, or yuck their yum. But as parents, we do want to have some sort of say in what they’re exposed to online before their brains have the capacity to process them. The risk to their mental health is real, and just YOLOing it hasn’t worked out too well.

    I’m sure there’s a lot of subtle behind-the-scenes stuff in the Colorado bill. I’ll wait to hear what EFF or Mike Masnick have to say about it. But as a techie, app developer, and parent, it reads like the least-worst way to keep a minor away from nasty crap without requiring every one of us to scan our faces and provide IDs to every rando website.


  • I’ve been a longtime mobile and web developer, have a teenage kid with a phone, and am a big privacy advocate (card-carrying member of ACLU and EFF). As a parent, I don’t want my kid exposed to cyber-bullying, toxic social media, or algorithmic bullshit.

    And I will tell you this: the operating system is 100% where you want to do age verification.

    I don’t want individual social media sites, dodgy third-party orgs, or government agencies scanning our faces or IDs. Under a family sharing plan, the OS already knows how old the kid is. Any site wanting to gate access can privately ask the OS if age > X without spilling their PII. Same concept as OAuth. An opaque, encrypted token indicating GO or NO-GO.

    Raging that they shouldn’t do any of this is just idiotic. Unfettered access got us CSAM, kids getting radicalized, or bullied to the point of self-harm. Fuck that.

    From a technical point of view, having OS-level verification is the least worst, and in my technical opinion, the best option.