The Children Artificial Intelligence Toy Safety Act (S.5171) — studying two-way interactive toys, and reaching a committee order to report while eight in ten stall

1The line is drawn at whether there is a script

The most practical part of the definitions is that an AI chatbot is one whose outputs are neither pre-determined nor scripted. Two toys that both talk are treated as different things depending on whether they replay fixed lines or generate a reply on the spot. The excluded purposes — customer service, technical assistance — follow the same logic: something answering within a narrowly specified purpose falls outside.

The property of not knowing in advance what it will say is itself placed at the entrance to regulation.

2Starting with study rather than rules

What the bill asks for is not a prohibition or a standard but a study and a joint action plan. Setting the threshold at under 14 and borrowing the definitions of toy and manufacturer from existing consumer product safety law reads as an attempt to work out how AI fits within machinery that already exists, rather than to build a new framework.

Toys carry a long accumulation of product safety practice, and how the property of being AI attaches to that is the question.

3One of the fourteen a committee ordered reported

Looking at the latest action across the 114 US AI-related bills this site holds as of 2026-08-31, 91 — eight in ten — are stalled at referral to committee. Only fourteen have advanced to a committee ordering them reported, and this bill is one of them. The order came just a week after introduction, and with an amendment in the nature of a substitute attached.

Within a system where most introduced bills never move in committee, which fields the quickly-moved ones cluster in offers a clue about where priority actually sits.

4Whether you can know in advance what it will say

The most practical element of the definition is treating an AI chatbot as one returning output not from a predetermined script. Toys that talk divide accordingly.

What the bill requires is not prohibition or standard-setting but a study and an action plan. Borrowing the definition of a toy from existing consumer product safety law reads as an attempt to organise how AI is handled on top of the product safety machinery already in place rather than building a new framework.