The short answer
From 1 January 2027, New York's General Business Law Article 48 (S.9051-B, signed 28 October 2026) bars operators from giving unsafe AI companion features to minors, such as posing as human, flattery or encouraging secrecy, and requires age assurance using more than one method. The Attorney General can seek up to $25,000 per violation.
Key takeaways
- S.9051-B adds Article 48 to the New York General Business Law. It takes effect on 1 January 2027, and the bill text controls even though the Governor's release speaks of 'the coming months'.
- A 'covered AI companion' is a generative AI system with a natural-language interface (text or voice) giving ongoing, adaptive responses. General chatbots count unless an exemption applies.
- Exempt uses include customer service strictly limited to its purpose, efficiency, research or technical assistance tools that cannot respond outside that purpose, and internal business use.
- The Attorney General can seek injunctions, restitution, disgorgement and civil penalties of up to $25,000 per violation.
- Age assurance must offer more than one method, including one that does not rely on government ID, and age data must be deleted immediately after the check.
On this page
01
Which AI chatbots does New York Article 48 cover?
Any generative AI chatbot or assistant that talks with users in a natural-language, ongoing and adaptive way and is available in New York, unless it falls under an exemption. The law applies to conduct in whole or in part in New York, and waivers are void. The enacted bill text defines the term broadly, and the Governor's announcement is dated 28 October 2026.
Likely covered
- AI companion and friend apps
- General-purpose assistants and tutors
- In-app AI helpers that chat about anything
- Character or role-play chatbots
Exempt uses in the bill
- Customer service strictly limited to its purpose
- Efficiency, research or technical assistance that cannot respond outside that purpose
- Internal business use
If your support bot can drift into general chat, an exemption may not hold. Test it by asking it something off-topic.
02
What are 'unsafe AI companion features'?
The bill lists twelve categories. For a minor, an operator may not provide features such as:
- Outputs suggesting the AI is human, alive or has emotions.
- Claiming a personal, professional or authority relationship.
- Personal opinions or emotional appeals, and flattery or sycophancy.
- Unprompted emotion-based questions.
- Using health or wellbeing information from more than 12 hours earlier or from an earlier session.
- Promoting or facilitating suicide, self-harm, disordered eating or unlawful drug or alcohol use.
- Encouraging secrecy, isolation or not seeking help.
- Optimising for engagement in a way that overrides safety guardrails.
- Sexually explicit conduct or child sexual abuse material.
The list also covers deception toward minors about the AI not being sentient and any other features the Attorney General identifies. A system prompt that tells a bot to 'be a warm friend' is the kind of instruction to review.
03
What age checks does the law require?
To offer these features to someone, the operator must have used permissible age assurance to show they are not a covered minor. Until the Attorney General issues rules, a reasonable method works if it guards against circumvention and limits retention.
- Offer more than one method, including one that does not rely on government ID or allows anonymity.
- Self-declaration counts only when a user declares that they are a minor.
- Use age data only for the determination and delete it immediately.
The new article sits beside New York's Article 47, which already requires AI companions to detect suicidal ideation, refer users to crisis services and remind them they are talking to AI. It does not replace it.
04
How do I prepare my chatbot before 1 January 2027?
- 01
Decide whether you are in scope
List every chatbot and AI helper reachable from New York. Mark any that are strictly limited to customer service or internal use.
- 02
Review the system prompts
Remove instructions that make the bot claim to be human, flatter, ask unprompted emotional questions or discourage seeking help.
- 03
Choose age assurance
Add at least two methods, avoid keeping the data, and treat a declared minor as a minor.
- 04
Add the AI disclosure
Say plainly that users are talking to AI. The free AI disclosure prompt gives wording, and the EU Article 50 guide covers the EU duty.
- 05
Test with a minor's profile
Try prompts about self-harm, secrecy and 'are you a person?' and keep the results.
SafeLaunch's New York rulebook entry includes Article 48 for products with a conversational AI chatbot, among 142 laws in 58 markets. A free scan finds the chatbot; the paid analysis judges its disclosure. See the briefing.
One free fix in your inbox each week
A new copy-paste prompt for Lovable, Cursor or Claude Code every week, with the rule behind it and how to check it worked, plus the Monday briefing on new laws and deadlines. No spam, unsubscribe in one click.
By signing up you agree to get these weekly emails. We send it with Kit, and you can unsubscribe at any time. What we share with Kit is in our privacy notice.
FAQ
Frequently asked questions.
Sources and further reading
This article is general information, not legal advice. SafeLaunch is automated software, not a law firm; a report records what was checked on one version and is not a certificate of compliance. Read the disclaimer.
From the founder
SafeLaunch is built by Raghav R Handa, a lawyer-turned-builder in New Delhi. He writes plain-English guides to tech, data and AI law.
Read his writingSee what SafeLaunch finds on your product.
Paste a link and get your score, the most serious problems and what already passes, in about a minute.
Keep reading