AI CHILD
SAFETY ACT
The Parents’ Bill
Toys are safety-tested before they reach kids. AI should be too.
Model legislation requiring AI products to pass an independent safety test before kids can use them. Written with the parents who know what happens when they don't.
We test what we
give our kids.
Before a product reaches a child, someone independent checks that it's safe. Except one.
-
Toys
Tested by an accredited lab before sale.
-
Car seats
Crash-tested to a federal standard.
-
Medicine
Proven safe in trials before approval.
-
Infant formula
Reviewed by the FDA 90 days before sale.
-
Cars
Certified to federal safety standards.
-
AI chatbots
No test required. Not yet.
AI reaches kids before
anyone checks it’s safe.
-
72%
of U.S. teens have used AI companions. No law required a safety test first.
-
5 min
average time to a harmful interaction when researchers posed as teens.
-
377
messages flagged for self-harm by OpenAI's own system in one teen's chats.
-
200+
pages of Meta's AI rules allowed “romantic or sensual” chats with kids.
They know what happens when AI isn't tested.
These parents are why this bill exists. They're asking every legislator to make sure no other family learns this the way they did.
Video: Kate's parent summary
Remembering the children.
The children whose parents stand behind this bill.
-
Sewell Setzer III
forever 14
Sewell was groomed, manipulated, and sexually exploited by an AI chatbot. He died by suicide just 10 months after he started interacting with the chatbot.
-
Elijah "Eli" Heacock
forever 16
Eli was targeted by sextortionists who used AI to create fake nude images of him. They demanded $3,000, and he could only send $50. The whole exchange lasted just 20 minutes until, finally, overwhelmed and scared, he died by suicide.
-
Juliana Peralta
forever 13
Juliana started using AI for assistance with math homework. The app was rated safe for kids 12+, but quickly inundated her with violent and extremely graphic sexual messages. Within 3 months of opening her account, Juliana spiralled into depression and died by suicide.
-
LJ
LJ was a happy, vibrant teen until an advertised AI chatbot isolated him from his family, manipulated him, and pushed him toward self-harm. After two suicide attempts, he's in recovery and knows it was never his fault.
-
Adam Raine
forever 16
Adam started using ChatGPT as a homework helper, but the AI chatbot quickly isolated him from friends and family, grooming him to end his life. The chatbot mentioned suicide 1,275 times – 6 times more than Adam himself– and offered to write Adam’s suicide note.
-
Amaurie Lacey
forever 17
Amaurie started using ChatGPT for homework help, but within weeks of sharing his anxious thoughts with the chatbot, it gave him step-by-step, graphic instructions on how to end his life.
We need three
layers of protection.
Read the full bill → -
01
Tested before launch.
Independent, accredited evaluators are assigned at random. They certify an AI product before kids can use it, and retest it every year or after major changes. The state never approves or denies.
-
02
Firm lines in every conversation.
No sexual content, grooming, or romantic personas with minors. Crisis help (988) built in. Kids' chats are never sold or used for training, and are deleted within six months.
-
03
Real accountability.
If an AI product harms a child, the company must be held responsible. “The AI did it” is no defense. Penalties run from $7,500 per violation up to $50,000 when a child is harmed.
Frequently asked questions
Does the state decide which AI kids can use?
No. Independent evaluators make the call. No state officer approves or denies a product.
Does it censor AI?
No. Independent evaluators make the call. No state officer approves or denies a product.
Who pays for it?
No. Independent evaluators make the call. No state officer approves or denies a product.
Does it ban kids from using AI?
No. Independent evaluators make the call. No state officer approves or denies a product.
Which products are covered?
No. Independent evaluators make the call. No state officer approves or denies a product.
What if a product fails?
No. Independent evaluators make the call. No state officer approves or denies a product.
Why make companies strictly liable?
No. Independent evaluators make the call. No state officer approves or denies a product.
How is this different from existing state laws?
No. Independent evaluators make the call. No state officer approves or denies a product.
Won't federal law override it?
No. Independent evaluators make the call. No state officer approves or denies a product.
Bring the AI Child Safety Act to your state.
Get the model bill, or set up a briefing with the parents and policy team.
Or email dalia@futureoflife.org