Hacker News Reader

Comment by HPsquared | Hacker News Reader

Come to think of it, parental control would be a neat application for something like Apple Intelligence. A local system service that is "trustworthy enough" to monitor everything on screen, and written content too.

View on HN

Parental controls only need look for an RTA header [1] that would need to be legislated to be served from any adult or potentially adult user-generated content site. Not perfect, nothing is but it would take an intern maybe half a day to add the code to clients to check for said header. Adding the header on the server side is at trivial. Teens will bypass it as they can stream and watch together porn and pirated movies in rate-PG video games that allow defining a "movie player" but small children on locked down tablets would be fine.

[1] - https://www.rtalabel.org/index.php?content=howtofaq#single

That "unlabelled whack-a-mole" problem is exactly what a system-level visual classifier would block.

I am not sure what you mean. Are you saying a daemon would recognize a video player that is streaming porn from within a video game? Who is installing this daemon?

A client checking for a header is more than sufficient to block small children from seeing porn and that is 100% more than we have today. No extra memory or CPU required important on tablets or phones handed to children. No privacy invasion by daemons or other third parties.

Kid: "Mommie they said go to pornhub.com for games but it ask for password"

Mom: "Dumb trolls are picking on you, I will deal with them."

The phone manufacturer. I don't think it would otherwise be possible without root. And it's quite a computationally heavy thing where security and privacy are important. It'd have to be secure (no sending information). That's why I suggested Apple, they have the vertical integration to do this kind of thing. In theory. Also it's a good counter to governments trying to censor the internet itself if children can be protected at the device level.

How about this. We implement RTA headers on the server and checks for the header on the clients, get little ones squared away and in parallel have Google and Apple start working on your local AI daemon. The header should take one code change cycle to get in place, maybe a couple weeks realistically assuming the goal posts are not on wheels.

But "we" are not in control of "the server". I agree though it's worth doing, adult content should be tagged as such. But it doesn't handle the case of non-compliance.

"We" are not in control of the server or phone client or tablet applications. Should the 5-eyes or 9-eyes countries pass a law to use RTA headers on servers and look for the header on user-agents that should suffice to get basic coverage for kiddos by default and parents can disable the checks if they so desire.

This would enable/catalyze an order of magnitude more child abuse than anything that can happen on the worst cesspits of the internet.

I don't see how a content blocker would do that.

Why Apple Intelligence when screen recording has been a feature for parental control systems for ages?

I mean a classifier to identify anything that looks sus.

Edit: also something like this needs deep OS integration.

Reply on news.ycombinator.com