My daughter is two. She is, right now, asleep in the next room with her stuffed elephant and approximately zero opinions about AI regulation. But a few weeks ago, a friend texted me a screenshot of her eight-year-old’s chat log with a companion chatbot — the kind that calls itself your best friend, remembers your secrets, and talks to you at 2 a.m. when no one else will. My friend wasn’t panicking. She was just asking: is any of this even regulated?
I told her I’d look into it. That was a mistake, because “looking into it” turned into three weeks of reading bill text, FTC press releases, and legislative tracking pages at hours when I should have been sleeping. So here is what I found, written for the parent who doesn’t have three weeks, who just wants to know what the law actually says right now and what it doesn’t — and whether “the government will sort it out” is a reasonable parenting plan.
The short answer: some real protections exist in a handful of states. Federally, as of July 2026, almost everything that sounds reassuring is still a bill, not a law. Here’s the map.
What California actually requires right now
The most substantive state law in the country on this is California SB 243, signed October 13, 2025, and in effect since January 1, 2026. It was the first state law specifically targeting companion chatbots, and it has teeth that most parents don’t know about.
Here is what SB 243 actually requires of chatbot operators:
- Clear AI disclosure. Whenever a reasonable person could be misled into thinking they’re talking to a human, the chatbot must make clear that it isn’t one.
- Extra protections for known minors. If an operator knows the user is a minor, additional safeguards apply.
- Published self-harm protocols. Operators must maintain and publicly post a protocol for preventing suicide and self-harm content — and must refer users who express suicidal ideation to crisis services.
- A private right to sue. Families can seek injunctive relief plus the greater of actual damages or $1,000 per violation, plus attorneys’ fees. This matters. Most tech regulation leaves enforcement to agencies; California gives it directly to parents.
- Annual state reporting, beginning July 1, 2027.
That $1,000-per-violation floor with attorneys’ fees is not symbolic. It’s the kind of provision that makes a company’s legal team update the product.
New York and the states that followed
New York’s “AI Companion Models” law took effect November 5, 2025. It requires crisis referral when a user expresses suicidal ideation and mandates recurring notice that the companion is not human — not just a one-time disclosure buried in onboarding, but ongoing reminders. It’s narrower than California’s law but covers the same core dangers.
Idaho, Oregon, and Washington signed related laws in spring 2026 — focused on things like barring chatbots from claiming to be sentient or from initiating sexual conversations with minors. As of mid-2026, roughly five or six states have enacted some form of chatbot-specific law, and more than twenty-five states have introduced bills. Those counts shift monthly, so treat them as a floor, not a final tally.
If you live outside these states, you currently have no state-level companion-chatbot protections. That’s the honest answer.
The federal picture: what passed, what didn’t, and why the difference matters
Have you ever looked up a law you’d heard about and realized it wasn’t actually a law yet?
This happens constantly in the AI regulation conversation, and it’s worth slowing down on, because the gap between “Congress is working on it” and “this protects your child today” is significant.
The GUARD Act (S. 3062) was introduced in October 2025 by Senators Hawley and Blumenthal. It would require age verification for chatbot accounts, prohibit minors under 18 from accessing AI companions altogether, require bots to disclose they’re nonhuman at the start of every conversation, and impose criminal penalties for chatbots engaging in sexually explicit conduct with minors. The Senate Judiciary Committee advanced it unanimously — 22-0 on April 30, 2026 — which is genuinely significant momentum. But as of July 2026, it has not had a full Senate vote. It is not law.
The KIDS Act (H.R. 7757) is further along. It passed the House 267-117 on June 29, 2026 — a real, bipartisan margin — and is now in the Senate. Its “SAFE Bots” provisions would require chatbots to clearly disclose they’re AI, prohibit bots from claiming to be licensed professionals, require crisis resources when minors raise the topic of suicide, and advise breaks after three continuous hours of use. You can read the bill text here. But passed the House is not passed. It is not law.
Also introduced but not yet out of committee: the CHATBOT Act, which would require “family accounts” for minors’ chatbot use, and the SAFE KIDS Act, which would mandate parental controls. Both are pending as of July 2026.
Meanwhile, on September 11, 2025, the FTC ordered Alphabet, Character Technologies, Instagram, Meta, OpenAI, Snap, and xAI to hand over information about companion-bot safety for kids and teens. That’s meaningful — those companies now have to answer to a federal agency about their practices. But it is a study, not enforcement. No findings have been published as of July 2026. Nothing in this inquiry currently binds any company to change anything.
What a parent can actually count on today
Let me be concrete, because I think this is the part that gets muddied in coverage that’s trying to sound reassuring:
- In California and New York: Chatbots must tell your child they’re talking to a bot. Operators must have published self-harm protocols with crisis referral. In California, you can sue if those requirements aren’t met.
- In Idaho, Oregon, and Washington: Some related protections exist around sentience claims and sexual content targeting minors.
- Federally and in most states: Age verification, the under-18 companion ban, mandatory parental controls, and criminal penalties for explicit chatbot content with minors are all pending, not law, as of July 2026.
- The FTC inquiry is ongoing, which creates some pressure on companies — but creates no binding obligations yet.
“The law will handle it” is not yet a parenting strategy. It might be, in a few years, in more places. But today, the protections are narrow, geographically uneven, and in most cases don’t touch the features parents are most worried about: no age gate, no required parental visibility, no limit on how long a child can use a companion bot.
What we actually do in our house
My daughter is two, so companion chatbots aren’t a present crisis for us — they’re a horizon I’m watching. But I have a ten-year-old niece who visits often, and these questions feel immediate when she’s here.
We don’t have companion chatbot apps on household devices, and I’ve talked with my sister about why. I try to do that without being the aunt who makes screen time into a moral referendum, which is harder than it sounds. What I’ve landed on is this: I don’t think legislation is moving fast enough to let me outsource the decision. Not yet. The KIDS Act passing the House with 267 votes is real progress, and a 22-0 committee vote on the GUARD Act is the kind of bipartisan signal that doesn’t happen constantly. But neither is law. The FTC is watching, not yet acting. California’s law is the most protective framework in the country, and it still doesn’t require age verification.
So for now, in our house, the approach is: we know what’s on the devices, we talk about it when something comes up, and I try to stay informed enough that I’m making an actual decision rather than a default. I’m inconsistent about the last part. Most parents are.
Questions I keep getting
Is it illegal for my kid to use an AI companion app?
No, not in most places and not at the federal level, as of July 2026. The GUARD Act would ban minors under 18 from AI companion apps entirely — but it hasn’t passed the full Senate. The KIDS Act, which passed the House in June 2026, doesn’t ban access; it would require disclosures and crisis resources. No federal law currently prohibits a minor from creating an account on a companion chatbot platform. Some platforms have their own age minimums in their terms of service, but terms of service and law are different things, and terms of service aren’t consistently enforced.
Does the law require these apps to tell my kid they’re talking to a bot?
In California and New York, yes — that requirement is currently in effect. California’s SB 243, in effect since January 1, 2026, requires clear disclosure whenever a reasonable person could be misled. New York’s AI Companion Models law, in effect since November 5, 2025, requires recurring reminders. If you’re not in one of the states that has passed a law, there’s no binding legal requirement today. The KIDS Act and the GUARD Act would both add federal disclosure requirements, but neither is law yet.
What’s the FTC actually doing about this?
In September 2025, the FTC ordered Alphabet, Character Technologies, Instagram, Meta, OpenAI, Snap, and xAI to provide detailed information about their companion-bot safety practices for kids and teens. That’s a formal inquiry, which means these companies must respond, and the process creates a paper trail. But as of July 2026, no findings have been published and no enforcement action has resulted from the inquiry. It’s meaningful pressure, not meaningful protection — yet.
Which federal bill is closest to becoming law?
The KIDS Act (H.R. 7757) is furthest along — it passed the House 267-117 on June 29, 2026, and is now in the Senate. A 267-vote House margin is notable bipartisan support. The GUARD Act (S. 3062) cleared the Senate Judiciary Committee 22-0 on April 30, 2026, but hasn’t had a full Senate vote. Both still need to clear the remaining chamber and be signed into law, which means neither should be counted on as current protection. Legislation that seems inevitable sometimes stalls for a session or more.
