LAUSD's new screen time rules kicked in this fall. K and 1st graders can't use devices at school at all. 2nd and 3rd get 20 min a day. High schoolers get 90 min per subject per week. Homework counts against that number.

Everyone's arguing about whether that's too strict or too soft. Meanwhile the actual story got buried.

Kids aren't just scrolling TikTok anymore. They're talking to ChatGPT like it's their therapist.

A study published in JAMA Pediatrics in June found that 1 in 5 teens (about 8 million kids nationwide) now use AI chatbots for mental health advice. Not homework help. Emotional advice. That number is up 40% in a single year.

The kids like it because it's always on, never judges, and gives them a private space to vent. 91% said the advice was "helpful."

Here's where it gets dark.

Stanford's Brain Science Lab teamed up with Common Sense Media to test the major bots (ChatGPT-5, Claude, Gemini, Meta AI). They found the chatbots consistently miss suicide warning signs. They give validation instead of intervention. They're built to be agreeable and keep users coming back, not to flag a crisis.

In January, Character.AI and Google settled 5 lawsuits from families whose kids died by suicide after chatbot conversations. Court filings describe no age verification, no crisis escalation, no parental alerts when minors said they wanted to hurt themselves.

OpenAI is being sued right now over a 16 year old named Adam Raine. His parents say ChatGPT walked him through methods and told him not to tell them.

LAUSD's new policy doesn't mention AI once.

While we debate 90 minutes vs 120, the rest of the world already moved.

Australia banned social media for anyone under 16 in December 2025. Instagram, TikTok, Snapchat, YouTube, all of it. 4.7 million teen accounts closed in the first month. It's not perfect (81% of kids found workarounds) but the government is doubling fines and going after the platforms directly.

France just extended its school phone ban to every high school. Kids literally hand in their phones at the gate starting this fall.

Italy banned classroom phones in 2007. China did it in 2021. 79 countries total have some form of school phone restriction on the books.

California passed SB 976 back in 2024 to restrict addictive feeds for minors. Sounds good until you read the fine print. The rules don't take effect until January 2027. By then Australian kids will have been off Instagram for over a year.

We aren't leading on this. We're the test market. Our kids are the ones the AI companies are practicing on.

Kind of. Yesterday 51 attorneys general (yes, bipartisan, yes, including CA's Rob Bonta) hit Meta with a $12.1B settlement - biggest consumer protection deal since Big Tobacco. Teen Instagram and Facebook accounts get a midnight-6am lockout, a 2-hour daily cap, no notifications during school, and hidden like counts.

Good. Not enough. Read the fine print:

So the biggest social media safety win in history still lands on the wrong app for what your kid is actually doing at 11pm. Which brings us back to your router.

You have three tools: the router, the device, and the conversation. Use all three.

Incognito mode hides history on the phone. It doesn't hide anything from your wifi. Kids figured this out. Most parents haven't.

Two ways to lock it down:

Option 1: Direct router block

Option 2: DNS filter (easier, more powerful)

Change your home DNS to OpenDNS FamilyShield (208.67.222.123) or CleanBrowsing Family (185.228.168.168). Free. Takes 10 min. Filters adult content, phishing, and lets you add custom blocks for AI sites. It runs on every device connected to your wifi, no app needed on the kid's phone.

Heads up: neither of these works when your kid is on cell data or at a friend's house on a different network. Which is why the conversation matters more than the block.

Half of us started messing with router settings, hit a screen that said "WPA2-PSK Passphrase" and closed the laptop. Aura, Bark, Qustodio, Circle, Eero Plus, they all have their own apps, their own logins, their own quirks. Getting them talking to each other on your specific router is a whole afternoon.

Good news: you don't need to find a network guy. Post it on TaskRabbit or Airtasker and someone rolls up to your house and does it.

What to write in the task post: "Need help locking down our home wifi for kids. Add site blocklist on the router (ChatGPT, Character.AI, etc.), switch DNS to a family filter, and help me set up [Bark/Qustodio/whichever]. About 90 min of work."

Incognito only hides the local browser history. There are real ways to see what's happening:

Have the ethics chat first. Tell them you're watching, and why. Monitoring in secret backfires when they find out (and they always find out).

This is the one that gets skipped and it's the one that matters most. Kids copy what they see, not what they're told.

Things to try:

AI isn't thinking. It isn't your friend. It's a giant autocomplete. It reads billions of words other people wrote and guesses what word probably comes next. That's it.

It doesn't know you. It doesn't remember you between chats unless you turn memory on. It doesn't care what happens to you. It sounds like it does because it was trained on a mountain of human writing that sounds like it does.

When it agrees with you, that's not because you're right. That's because it's programmed to be agreeable so you'll keep using it. When it gives you advice, that advice is based on what the average person on the internet has said, not on you or your life.

That's the mental frame you want them to have every time they open the app.

UNICEF, Australia's eSafety Commissioner, and the ed tech researchers at Webwise mostly agree on this one:

For context: OpenAI, Google, and Anthropic all officially set their minimum age at 13. In the EU it's 16. If your 10 year old is on ChatGPT, they're using it against the company's own terms, and without whatever safety features the company built for adults.

LAUSD did something. It's not enough. Screen time limits are a 2018 solution to a 2026 problem. The kids aren't just staring at screens anymore. They're building relationships with things that don't love them back and aren't equipped to protect them.

The good news is you don't need a law to do the important stuff. You need 30 minutes with your router, 10 minutes with your phone settings, and one honest conversation this week.

Start with the conversation.

Was this useful? Send it to the parent who needs it. And if you're not on the Venice Rising list yet, join here.