New York's two new child-safety laws get the basics right

New York Governor Kathy Hochul has signed a law meant to protect children from AI chatbot harms, and on the same day she signed a second law requiring connected baby monitors sold in the state to block unauthorized viewers and listeners.¹ Two bills, one idea, and I think it is a good one.

Here is my argument in one sentence: laws that name specific, checkable behaviors protect families better than any company’s promise to do better.

The chatbot law is specific. AI companies must verify users’ ages, and they must offer at least one method that does not rest entirely on a government ID. They must delete that age information after use. If a user is under 18, or has not yet been confirmed as an adult, the chatbot cannot pretend to be a real person or to have human emotions. It cannot claim a romantic or authority-figure relationship with a child, push unprompted emotional questions, or nudge a kid to keep secrets or avoid asking trusted adults for help.¹

The baby monitor law is the humbler sibling, and I like it for that. Connected monitors must stop unauthorized users from seeing or hearing what the device captures, and they must carry warning labels explaining why security features matter.¹ That is a floor, not a ceiling. Floors are what buyers need, because nobody can inspect a camera’s firmware at the store.

Now the best counterpoint, stated fairly. Age verification means collecting something about every user, adults included, and a rule that sorts people into “confirmed adult” and “everyone else” will annoy plenty of grown-ups who just want to ask a chatbot a question. Rules written today may also fit awkwardly onto products that change every few months. Those are real costs, and a skeptic is right to name them.

I weigh them differently, for two reasons. The law answers the data worry directly: age information must be deleted after use, and companies can face civil penalties of up to $25,000 per violation, plus potential deletion of unlawfully collected data and the models trained on it.¹ That is a bite, not a warning label. And the banned behaviors are not exotic edge cases. Governor Hochul herself said she was not sure why a law was needed against something so obvious.¹ When the list of prohibited conduct is that plain, the burden of proof belongs with the company that wants to do it anyway.

The scope matters for tech buyers too. The chatbot law takes effect in six months and the baby monitor law in 90 days, so manufacturers and AI companies have a defined window to adjust.¹ Parents get a complaint website that goes to the state attorney general, which turns a vague unease into a filing.¹

What would change my mind? If age checks turn out to push everyday users toward weaker, less private tools, or if the baby monitor rules produce labels nobody reads and security nobody tests, I would say the laws need rewriting. Results, not signing ceremonies, are the measure. I would also want to see how enforcement actually goes once complaints start arriving.

Our read: Common Sense Media’s founder called the chatbot law a template for the rest of the country,¹ and other states are already moving against AI companies in court. Whether that template spreads depends on a dull question: do the rules work once they are real? Watch the first six months.

A good safety law is boring on purpose: it says what you cannot do, and it says it clearly.

Footnotes

1. Engadget, “New York governor signs AI chatbot child safety law”. Retrieved October 28, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸