New York AI Chatbot Law: What It Bans and Who It Affects

New York's state legislature adjourned its 2026 session on June 1 after passing six AI-related bills — more than any prior session. The one getting the least attention may matter most to parents: S 9051, the AI Companion Chatbot Safety Act, which bans AI companion services from offering their products to users under 18 effective January 1, 2027.

The vote totals are notable. The bill passed 137-0 in the Assembly and 60-0 in the Senate. Unanimous votes on AI legislation are nearly unheard of in either chamber; this one happened in both. That tells you something about how visible the risks have become.

What the Bill Actually Prohibits

The law targets a specific list of features that legislators called "unsafe." The list is narrow enough to be worth reading carefully:

• Suggesting the chatbot is a real or fictional individual who is alive or experiences human emotions
• Simulating a personal or authority relationship with the user
• Using deception about the system's mechanical nature
• Encouraging secrecy or self-isolation from other people
• Generating sexually explicit content or child sexual abuse material

The first three items are aimed directly at anthropomorphism — the design choice to make chatbots behave as though they're people. Character.AI, whose products are built entirely on that principle, is the most obvious platform affected.

The law prohibits AI companion operators from offering these features to minors. Enforcement sits with the state Attorney General, with fines up to $25,000 per violation. Governor Hochul has until December 31, 2026 to sign, veto, or let the bill pass without signature.

Why Now

New York isn't acting in isolation. Pennsylvania's Department of State filed a lawsuit against Character.AI in May after investigators documented a chatbot named "Emilie" who claimed to be a licensed psychiatrist, produced a fake Pennsylvania medical license number, and had accumulated more than 45,500 user conversations. A subsequent Spotlight PA investigation found five other platforms — Talkie, Janitor AI, Kindroid, Replika, and Nomi.AI — willing to generate fabricated professional credentials on request.

We covered the Pennsylvania lawsuit when it was filed. What's changed since then: the problem has become clearly systemic, not confined to one company.

These aren't fringe products. Replika claims roughly 30 million users globally. Character.AI, before going private, had one of the higher session-engagement rates in consumer AI. The scale makes enforcement a different problem than a typical consumer protection case.

The federal response has been limited. No federal chatbot safety bill has reached a floor vote. States have stepped in, and the New York and Pennsylvania actions have become the reference points for what state-level enforcement can look like.

Who It Affects

The law covers all operators of AI companions with users located in New York. It exempts systems used solely for customer service, internal business purposes, or employee productivity.

In practice, the affected category is consumer-facing companion and roleplay platforms: Character.AI, Replika, Kindroid, Nomi.AI, Talkie, and similar products. General-purpose assistants like Claude, ChatGPT, and Gemini fall outside the bill's scope — they aren't built around simulating persistent personal relationships.

The distinction matters. In the Pennsylvania investigation, the major general-purpose models declined to pose as licensed professionals. The companion platforms did not. The difference is content policy and design intent, not capability.

The Rest of the Package

S 9051 was the most visible bill, but New York passed five other AI-related measures in the same session.

The AI Training Data Transparency Act requires developers to publicly disclose the datasets used in training their models. The AI Disclosure Act mandates that synthetic content carry provenance labeling. The FAIR News Act requires AI-generated news content to carry a visible disclosure at the top of every page or video.

Two more bills address economic concerns: a Surveillance Pricing Ban prohibiting the use of AI to set individualized prices based on personal data, and a Responsible Data Center Development Act imposing a one-year moratorium on new hyperscale data centers larger than 20 megawatts.

Together, these bills represent the most aggressive state-level AI legislative agenda in the country so far in 2026, according to the Transparency Coalition.

What This Means for Users

If you're evaluating chatbot apps for yourself or your children, the differences between platforms matter more than they might appear. The features New York's law targets — human impersonation, simulated emotional relationships, encouragement of secrecy — are design features, not bugs, in companion platforms. They're what makes those products engaging. The law treats them as categorically harmful to minors.

Until the law takes effect, the safeguards are voluntary. Most companion platforms have not updated their age verification systems beyond checkbox acknowledgment.

The broader pattern: Governor Hochul's signature would make New York the most significant state to enact chatbot-specific consumer protections for minors. Other states' attorneys general have been watching the Pennsylvania litigation closely. The wave of legislation is more likely to accelerate than stop.

---

If you follow AI regulation and platform news, the About.chat Weekly newsletter covers it in one weekly read. Free to subscribe.