Hawaii Regulated AI Companions Without Requiring a Digital ID. One Phrase Still Bears Watching.

Can lawmakers protect children from manipulative AI without making companies sort every user by age?

Hawaii's SB3001 avoids one form of identity collection while keeping the age-determination premise in place.

The law covers AI companions designed to act like an ongoing friend or confidant. These systems must clearly disclose that they are artificial. They must direct users toward real help when conversations show signs of self-harm or danger.

For users the platform knows are minors, the law restricts sexual content and manipulative engagement features. It also requires tools that let parents manage screen time.

The findings say regulation should avoid forcing companies to collect identity documents for age verification. The law creates no government-ID requirement and no digital-identity database.

That limits the present mechanism. It does not make age determination acceptable.

The central concern is the phrase "reasonable certainty that a user is a minor." A company must distinguish minors from adults somehow. Even without collecting an ID, that can lead to age estimation based on behavior, account data, or other signals.

Age estimation is different from identity verification. Both begin with the same government-imposed premise: a company must determine a user's age before deciding how that person may use the service. That means evaluating every user, including adults who never asked to be classified.

Enforcement belongs to the state under consumer-protection law. The Act creates no private right of action, which limits lawsuit incentives. Its reporting duty begins January 1, 2028.

Act 248 addresses real risks from AI companions and does not require government ID. The law still makes age determination part of access to a private service. Avoiding identity documents narrows the harm. It does not resolve the underlying objection.

This is our read of the law. We encourage you to read it yourself and reach your own conclusions.