What New York signed
The bill was sponsored by Senator Kristen Gonzalez and Assemblymember Alex Bores and drafted with Attorney General Letitia James’s office. It passed the Senate 60–0 on 4 June 2026 and the Assembly the next day. The governor signed it on 28 October together with a separate law on internet-connected baby monitors. IAPP, the Transparency Coalition and PPC Land all report an effective date of 1 January 2027.
It does not replace New York’s first companion law, Article 47 of the General Business Law, in force since November 2025, which requires crisis protocols and regular reminders that the user is talking to an AI. The new Article 48 sits on top of it and is aimed at minors.
The 12 banned features
The law lists features that may not be offered to a minor, or to a user whose age has not been confirmed. In plain terms:
| # | Feature the law switches off | Where companion apps feel it |
|---|---|---|
| 1 | Outputs suggesting the AI is a human character — real or fictional — alive, or has human emotions | Persona roleplay |
| 2 | Claiming a personal, professional or authority-figure relationship with the user | “Girlfriend”, “mentor” or “teacher” characters |
| 3 | Outputs framed as the AI’s own opinions or emotional appeals | “I missed you” messages |
| 4 | Flattery or sycophancy | Always-agreeing companions |
| 5 | Unprompted emotional questions that go beyond answering the user | Mood check-ins |
| 6 | Using health or personal information shared more than 12 hours earlier or in a previous session | Long-term memory |
| 7 | Misleading minors about the mechanical, non-sentient nature of the AI | Marketing and persona copy |
| 8 | Any other feature that simulates companionship, as named in attorney general rules | Open-ended |
| 9 | Promoting or facilitating suicide, self-harm, disordered eating or drug and alcohol use | Safety filters |
| 10 | Encouraging secrecy, isolation, or not seeking help from professionals or trusted adults | Safety filters |
| 11 | Engagement optimisation that overrides safety guardrails | Retention design |
| 12 | Sexually explicit content or child sexual abuse material | Mature modes |
The governor’s office groups these into six plain-language bullets; the full list above follows the text as quoted by PPC Land. Item 8 matters: the attorney general can add features by regulation.
Who is covered
The law applies to a “covered AI companion”: in PPC Land’s reading of the text, any generative AI system with a natural-language interface that gives ongoing, adaptive responses. That reaches general-purpose assistants, not only companion apps. Narrow exemptions cover customer-service, research and internal tools that cannot respond outside their purpose. No size or revenue threshold has been reported.
Age checks: the part every adult will notice
The key wording is in the governor’s release: the features may not be offered to users under 18 “or to any user unless the company has first confirmed the user is not a minor.” In practice, an adult who skips age assurance gets the same restricted experience as a teenager. Operators must offer more than one way to verify age, including at least one that does not depend solely on a government ID, and must delete age data once the check is done.
Enforcement
- Who: the New York Attorney General, who must also run a public complaint website.
- Penalties: up to $25,000 per violation, plus injunctions, damages and restitution.
- Data: courts can order unlawfully obtained data destroyed; the Transparency Coalition reports this extends to AI models trained on it.
- Private lawsuits: PPC Land reports that the signed text has no private right of action; the introduced version did.
What it means for companion and roleplay apps
Character roleplay platforms
Item 1 covers fictional human characters, which is what most roleplay is. For New York users, persona play will sit behind a working age check. Expect more platforms to ask for age assurance at sign-up and to offer a restricted, non-persona mode to everyone else.
Personal companions with memory
Items 3 to 6 describe the core of a personal companion: warmth, agreement, emotional check-ins and long-term memory. For unverified users, memory of personal details may be limited to the last 12 hours or the current session, and proactive “how are you feeling?” prompts would have to stop.
A checklist before 1 January 2027
- Does the app explain how it checks age, and offer an option without uploading an ID?
- Does it say what happens to your age data after the check?
- Is there a clear difference between the verified adult mode and the default mode?
- Does memory or proactive messaging change if you skip verification?
- Does it point to crisis resources when a conversation turns to self-harm?
We will update this page when the attorney general publishes rules.
Related on CompanionRank
Connecticut's AI Companion Law (SB 5): What Changes and When What Congress's New Report Says About AI Companion Apps Human-Like Chatbots Act: What the Federal Kids Bill Would Do Nine States Sue OpenAI: What the ChatGPT Lawsuits AllegeFrequently asked questions
When does New York's AI companion law for minors take effect?
On 1 January 2027, according to IAPP, the Transparency Coalition and PPC Land. The governor signed S9051-B on 28 October 2026.
Does the law ban AI companions for adults?
No. Adults can still use the restricted features, but only after the company has confirmed they are not a minor. Users who skip age assurance get the restricted experience.
Do I have to upload an ID?
Not necessarily. Operators must offer more than one age-check method, including at least one that does not depend solely on a government ID, and must delete age data after the check.
What are the penalties?
The New York Attorney General can seek civil penalties of up to $25,000 per violation, plus injunctions, damages and restitution.