> if they have reason to believe users are underage

The law requires "clear and convincing information", not merely "reason to believe". And since the law requires developers to rely on the provide age signal as the primary indicator of the user's age, developers are not incentivized to create a system that uses sophisticated data mining to derive an estimated age. If someone posts a comment on a YouTube video saying "I'm twelve years old and what is this?", that would absolutely not require YouTube to immediately start treating that account as an under-13 account.

▲

heavyset_go 17 hours ago | parent [-]

That would have to be litigated in court, and the easiest and cheapest way to avoid litigation is to just scan faces and IDs so you're sure your users won't upload or say anything that can bankrupt you while you sleep.

▲

wtallis 17 hours ago | parent [-]

It would be at least as valid a strategy to avoid collecting any unnecessary personal information about your users, so that you don't have to worry about whether the information you've amassed adds up to "internal clear and convincing information".

Remember, only the state AG can bring a suit under this law, and the penalty is limited to $2500 per child for negligent violations. It's probably cheaper to get insurance against such a judgement than to implement an invasive ID-scanning age verification system (and assume the risks of handling such highly-sensitive personal information).

▲

heavyset_go 17 hours ago | parent [-]

No platform is going to forgo analytics and using demographic information for advertising, that's their bread and butter.

I'd also argue it's clear and convincing if a kid changes their profile picture to a selfie of themselves, says they're 12, says they're in grade school, etc. Any reasonable person would take that at face value.

> implement an invasive ID-scanning age verification system (and assume the risks of handling such highly-sensitive personal information)

It's already implemented as face and ID scans by all the major platforms as it is. The systems are already there and they're already deployed.

Apps and platforms already integrate with 3rd party age verification platforms who handle the face and ID data, nothing ever has to touch your servers.

▲

Dylan16807 17 hours ago | parent | next [-]

> I'd also argue it's clear and convincing if a kid changes their profile picture to a selfie of themselves, says they're 12, says they're in grade school, etc. Any reasonable person would take that at face value.

That's so fragile, and it's not like they're making those claims to the site, it's natural language posting.

And someone who knows what they're doing would never take "I'm twelve years old and what is this?" at face value.

▲

heavyset_go 16 hours ago | parent [-]

Would you be willing to send explicit content to someone who presents themselves as a child online? A reasonable person wouldn't.

No one is suggesting a meme should be taken literally.

▲

Dylan16807 16 hours ago | parent [-]

You've completely changed the scenario. A human doing a one on one examination and personally sending data is totally different from a website allowing an account to exist and browse.

▲

heavyset_go 10 hours ago | parent [-]

I'm using send as a synonym for serve, would you serve such content to someone who presents themselves as a child on your platform? No.

Like do you really think someone who presents themselves as a child on Pornhub will stay registered and not banned? They aren't going to serve porn to someone who presents themselves as a kid.

I think it's pretty clear and convincing when someone presents themselves as a child on your platform. I'd be convinced and wouldn't take that liability on.

	▲	hellojesus 6 hours ago \| parent \| next [-]
		I agree with you, but then the implication becomes, "any platform that allows user content must scan all content posted or uploaded by the user to determine if any of it suggests they are a minor". I adds a bit of burden to all sites under the threat of civil penalties. Say I own a site that allows user posts like HN. If someone mentions they're in grade school in any post and later starts receiving links to adult sites in their DMs, now I'm liable.
	▲	Dylan16807 4 hours ago \| parent \| prev [-]
		"presents themselves" is too vague here. Presents to who? When you talk about a person instead of a webserver, that person is actually reading the profile, which makes a big difference. In the pornhub situation, someone who reads the comments will report them. The website doesn't need to throw machine learning at everything people type.

▲

crote 16 hours ago | parent | prev [-]

... except that analyzing profile pictures isn't exactly reliable (plenty of people use photos of their cats), people lie in chat, and advertising profiles are at best an educated guess.

The current analytics profiles are closer to "definitely into Roblox, 70% chance of being 13-18" than "This user was beyond any reasonable doubt born on 07-03-2002". Calling them "clear and convincing information" would be a massive exaggeration.