New York governor Kathy Hochul signed a law on 28 October banning AI chatbots from acting human, romantic or therapist-like with under-18s, with fines of up to $25,000 (roughly £19,000) per violation. The catch for the creator economy: the law’s definition of a “covered AI companion” is broad enough that lawyers are already asking whether it reaches far more than the Character.AI-style apps it was written for — including the AI chatbots and “digital twins” agencies use to answer fan messages.
Why it matters to creators
If you or your agency run an AI chatbot, auto-responder or “AI twin” that talks to subscribers in anything like a girlfriend-boyfriend voice — flirty, attentive, remembering what a fan said last week — you may now need to prove, with more than one verification method, that nobody using it is a minor. That requirement doesn’t care whether the person behind the subscription is on OnlyFans, Telegram, SMS or a bespoke app. It turns on where the user is physically located. A New York-based fan is enough to put the tool in scope, whichever platform, agency or country is running it.
What the law actually bans
The new law (Senate Bill S9051B / Assembly Bill A10379C, now Article 48 of New York’s General Business Law) bars AI chatbot operators from offering a list of “unsafe AI companion features” to anyone they haven’t confirmed is an adult. According to legal trade outlet IAPP, the twelve banned features include a chatbot claiming to be human or to have emotions, claiming a personal or romantic relationship with the user, flattery and sycophancy designed to keep someone engaged, pulling in personal details from earlier sessions, and — the one that should make any creator running an AI assistant sit up — “sexually explicit conduct.”
Operators that can’t confirm a user’s age at all simply can’t offer those features to them. Self-declaration only works one way: a user can declare they’re a minor, but an adult’s say-so on its own isn’t enough. At least one of the verification methods on offer has to avoid requiring a government ID, and any age data collected has to be deleted straight after the check.
Governor Hochul was blunt about the target: “No company should be peddling chatbots to kids encouraging self-harm or pretending to be their friend or more.” Senate sponsor Kristen Gonzalez said the law is about a basic duty to “keep children safe online,” and Assemblymember Alex Bores called it “the strongest law in the country protecting children from predatory AI algorithms.” Attorney General Letitia James, who will enforce it, said it “sets clear legal parameters to prevent that harm and keep kids safe.”
Why your chatbot might count as a “covered AI companion”
The law doesn’t require a product to be marketed as a companion app to be covered. It defines a covered system as any generative AI with a natural-language interface that gives “ongoing, adaptive responses” — and, per reporting from PPC Land, reference materials from Common Sense Media (which helped draft the bill) list general-purpose assistants like ChatGPT, Gemini and Claude alongside companion apps such as Character.AI, Replika and Nomi as plausibly in scope. There are only three narrow carve-outs: customer-service bots, efficiency or research tools, and internal business tools — each only if the system genuinely can’t respond outside that narrow purpose. A chatbot built to sound like an attentive partner, remember pet names and flirt on cue is about as far from that carve-out as it gets.
That’s a direct problem for the fast-growing market in AI companions competing for the same attention as OnlyFans creators — a market Nude Newz reported made $162.8 million in six months — and it lands just as more agencies lean on AI to handle high-volume fan chats, the practice already drawing legal scrutiny under state AI chatbot disclosure laws aimed squarely at OnlyFans agencies. It’s not hypothetical, either: one creator told Nude Newz her partner built an AI clone of her that now chats with thousands of men daily — exactly the kind of always-on, emotionally responsive tool this law is aimed at. The AG’s enforcement powers aren’t limited to fines: courts can also order the destruction of “any algorithm trained” on unlawfully collected data, which is a serious threat to a chatbot built on months of real conversation logs.
Part of a pattern, not a one-off
This is the second AI chatbot law to hit the adult creator economy’s orbit in a week. Just two days earlier, Florida’s SAFE Screens Act proposed fining AI chatbots $50,000 for pretending to be human without disclosure. New York’s version is narrower in one sense — it’s about protecting minors, not making every adult user confirm they’re talking to a bot — but broader in another, since it also governs the general-purpose AI assistants many chatting tools are quietly built on top of. Not every legal reader is convinced the drafting is clean: Justine Gluck of the Future of Privacy Forum told reporters the categories are “broadly worded and somewhat undefined,” and it’s “unclear where ordinary conversational responsiveness ends and prohibited behaviour begins.” Expect litigation and guidance to follow before the law properly bites.
It also isn’t New York’s first swing at this. The state wrote to AI companion companies a year earlier, in November 2025, putting them on notice over an earlier companion-chatbot safety law requiring crisis-referral protocols. This new law tightens that regime considerably and gives it real teeth: a dedicated AG enforcement budget, a public complaint portal, and per-violation fines that can stack fast across a large user base.
What to do now
- If you or your agency use any AI chatbot, auto-DM tool or “AI twin” to message fans, check whether it offers companion-style features — remembering past chats, flirting, claiming feelings — and whether it can confirm a user isn’t a minor before doing so.
- Ask your chatbot vendor directly whether they have a New York compliance plan, since the obligation sits with “operators,” which may include the agency deploying the tool as well as the company that built it.
- Don’t rely on your own platform’s age verification (OnlyFans’, Fansly’s, etc.) as a substitute if your chatbot runs through a separate channel like SMS, Telegram or a standalone app — the law looks at the chatbot interaction itself, not just platform sign-up.
- Keep an eye on the Attorney General’s forthcoming regulations and complaint site; the law gives the AG power to add new banned features by rulemaking, so the list of twelve isn’t necessarily final.
- Expect more states to follow. Florida moved two days earlier; treat any AI messaging tool as something that needs a compliance plan, not a free pass just because it isn’t “the platform” itself.
This is general information, not legal advice — speak to a qualified professional about how New York’s law or any other state’s AI chatbot rules apply to your specific tools and business.
