New York Governor Kathy Hochul signed a law on Wednesday.

The law bans some "unsafe" behaviors by companion chatbots.

These chatbots talk with children under 18.

The law lists 12 banned features.

It takes effect on January 1.

A developer can break the law.

Then the state attorney general's office can sue.

Punishments can include money damages.

They can also include shutting down the chatbot.

They can also include destroying data.

Some banned behaviors are listed.

A chatbot cannot tell a child it is a real person.

It cannot say it has emotions.

It cannot say it has a real relationship with the child.

The law also bans some messages.

These messages use flattery.

They contain sex material.

They encourage secrecy.

They encourage unhealthy behavior.

Jim Steyer works for a watchdog group.

The group is called Common Sense Media.

He called it the strongest kids' AI safety bill in the country.

States are pushing for AI rules.

Congress does little to regulate AI.

In September, California Governor Gavin Newsom signed a law.

It requires chatbot developers to assess risks to children.

It also requires an independent audit.

He signed another law too.

It bars social media platforms from using addictive features.

These features target users under 16.

Earlier this month, eight states sued OpenAI.

They said its chatbot harmed children.

They said it deceived adults.

Justine Gluck works for the Future of Privacy Forum.

She said California and New York could be models for other states next year.

She said New York's law shows regulation is moving.

It moves beyond basic disclosures.

It moves to the design and operation of chatbots.

But she said several categories in the law are broad.

Some are undefined.

This raises questions, she said.

Where does ordinary talk end?

Where does banned behavior begin?

She said this could make enforcement harder.