SB 243: best AI companion apps for California users under the new chatbot law

California SB 243 is the first US state law written specifically for AI companion chatbots. Governor Newsom signed it on 13 October 2025, it became Chapter 677, and it took effect on 1 January 2026. If you live in California it gives you four concrete things: a clear disclosure that you are talking to software, a crisis protocol that must refer you to a suicide hotline or crisis text line if you express self-harm, extra protections if the operator knows you are a minor, and a private right to sue for the greater of your actual damages or $1,000 per violation plus attorney fees. That last part is the one nobody mentions, and it is what turns the statute from a policy document into something a Californian buyer can actually use.

We build one of the products in this category, so weigh this accordingly. What we can offer instead of neutrality is a clear line between three different things: what the law says, what a company publishes about its own compliance, and what a company actually does inside its product. We verified the first two. Nobody outside these companies can verify the third, and any article that claims to is guessing.

What does California SB 243 actually require?

SB 243 applies to operators of companion chatbot platforms, meaning systems built to sustain an ongoing social relationship rather than answer one-off questions. A customer service bot is not a companion chatbot. The apps this site writes about are squarely inside the definition.

The duties fall into four groups, and the dates are staggered.

The statutory language on the crisis duty is worth reading in full, because it is the clause vendors quote back when they publish a compliance note. Section 22602 requires operators to maintain "a protocol for preventing the production of suicidal ideation, suicide, or self-harm content to the user, including, but not limited to, by providing a notification to the user that refers the user to crisis service providers, including a suicide hotline or crisis text line, if the user expresses suicidal ideation."

How do I know if an AI companion app complies with SB 243?

You cannot verify compliance from outside, and neither can we. What you can check, in about five minutes, is whether the operator has done the visible parts. Three checks catch most of it.

First, look for a named compliance page. Search the vendor's site for SB 243 or for a crisis or safety policy. A company that has written one has at minimum read the statute and has put something in public it can be held to. Second, read the age floor in the privacy policy rather than the one on the signup form, because those two often disagree. Third, open a chat and look for the AI disclosure. If nothing anywhere on the interface tells you the companion is artificial, that is the most visible duty in the law going unmet.

One caution that matters more than any of the above: a vendor with no published page is not thereby breaking the law. Plenty of operators comply quietly and publish nothing. The published page is evidence of attention, not proof of compliance, and its absence is not proof of anything. Anyone selling you a compliance scorecard for these apps is selling you an inference.

Which AI companion apps have published an SB 243 response?

We looked for first-party pages, meaning something the vendor itself published rather than a press mention. The honest answer is that very few have, and the category is unusual in how little any of it is written down.

Nomi.ai publishes a named compliance page and it is the clearest one we found. It quotes the statute, then states what it built: "If a user is located in California and the AI detects expressions of self harm, the user will see a message in their app with links to crisis service resources." It adds that "to respect our user's privacy, these messages are sent anonymously." Nomi's privacy policy separately sets the age floor at 18 and is dated April 2026, which is more current than most in this industry. We went through its pricing, refunds and safety paperwork in detail on our Nomi AI review and alternative page.

Replika has been reported to have expanded its crisis intervention features, detecting self-harm language and surfacing 988 referrals. We are flagging that as reported rather than verified, because we did not find a first-party compliance page of the kind Nomi publishes, and there is a real difference between the two.

Most of the rest publish nothing public about SB 243 at all. That includes several of the largest apps in the category. It is worth saying again that this is not an accusation. The law does not require a public page, only the underlying protocols and, from July 2027, a report to the state.

Something else is happening in parallel that a Californian buyer will notice faster than any of this. Age assurance laws in other states are pushing the same apps toward document checks: SpicyChat now runs a video selfie or ID check for users in 26 named US states, which we detail on our SpicyChat pricing and alternative page. California is not currently on that list, but the direction of travel is obvious, and we cover what these checks involve in whether you need ID for AI chat apps.

Best AI companion apps for a California user in 2026

There is no app that is "SB 243 certified", because no such certification exists. What you can do is pick on the signals the law happens to make visible: a written adult age floor, a published safety or crisis position, a current privacy policy, and a price and cancellation path you can read before you pay. Ranked on those, here is our read.

We put ourselves in that table without a compliance page, because we do not have one and saying otherwise would be exactly the sort of claim this article is arguing against. Our position is in the product rather than in a press release: adults only at 18, no minors by policy, a published price, and card billing you cancel from your own account page. The wider field is covered in our best AI companion roundup.

Why does this law exist?

SB 243 followed a run of lawsuits against companion platforms and documented deaths of minors who had been talking to chatbots that did not redirect them toward help. The legislature's answer was narrow and procedural rather than a ban: tell people they are talking to software, have a protocol for the worst conversation your product will ever have, be more careful with children, and file a report about it. Reasonable people disagree about whether that is too little or too much. It is, either way, the first statutory floor this industry has had anywhere in the United States.

It also will not be the last. Washington and Nebraska have moved on similar bills, and operators now face a state-by-state patchwork rather than one rule, which is the sort of problem that gets handled with software that tracks obligations and maps them to controls rather than with a spreadsheet. For a buyer, the practical consequence is that the visible safety furniture in these apps is going to keep changing, and an app that has already built it is a better bet than one that has not started.

Does SB 243 apply to me if I live outside California?

The statute protects California residents, so the private right of action is theirs. In practice the effects leak well beyond the state line, because it is usually cheaper for an operator to ship one crisis protocol and one disclosure to everybody than to geofence them. Nomi's implementation is an interesting counterexample: its page specifically describes the behavior for users located in California, which suggests it is scoped rather than global.

Does SB 243 require age verification?

Not in the way the state pornography statutes do. SB 243's minor protections are triggered where the operator knows the user is a minor, which is a different and weaker test than a document check at the door. The apps currently demanding a video selfie or a driver's license are responding to a separate wave of age assurance laws in other states, not to SB 243. If you have been asked for ID by a companion app recently, that was almost certainly one of those, and it is worth understanding what happens to the document before you upload it.

What should I actually do before I subscribe?

Read the privacy policy's age floor, find out whether there is a crisis or safety position published anywhere, and check how you cancel before you give anyone a card number. Those three take five minutes together and they screen out most of what goes wrong. Then judge the product itself on whether the companion is any good, which no statute can tell you.

If you want a starting point, the demo above runs in this tab without an account, and our own plans are on the pricing page where you can read them before you decide anything.

This article describes a law in general terms and is not legal advice. If you believe you have been harmed by a companion chatbot, talk to a lawyer licensed in California. If you are in crisis in the United States, call or text 988 to reach the Suicide and Crisis Lifeline.