SAFE KIDS Act: AI Child Safety by Design

The American Council supports SB4855, the Safeguarding AI Features to Ensure Kids' Informed Digital Safety Act, and calls on every member of Congress to vote yes and move this legislation into law. AI chatbots are now a daily presence in the lives of American children, and federal law has not kept pace. This bill closes that gap in a principled, structured way that honors the dignity of children and the God-given responsibility of parents to protect them.

The bill's age verification framework is rigorous without being reckless. Providers must implement age estimation technology and treat every unverified user as a child until age is confirmed [Fact 3, 4]. When age signals conflict, the signal indicating the younger age controls [Fact 7]. Providers must periodically re-screen all accounts to maintain ongoing compliance [Fact 8]. Critically, any data collected for age estimation purposes is strictly ring-fenced: it cannot be retained beyond what is necessary, used for any other purpose, or sold to any third party [Fact 10]. We believe these protections reflect exactly the kind of precautionary ethic that a just society owes to children who cannot fully protect themselves.

The content and conduct prohibitions are direct and serious. Providers may not knowingly or recklessly make available to child users AI chatbots that generate content promoting eating disorders, graphic violence, sexually explicit material, CSAM, or suicidal ideation [Fact 14]. The bill goes further, prohibiting AI chatbots that engage in sexually explicit conduct with a child or solicit the creation of sexualized depictions of children [Fact 21]. These are not administrative technicalities; they are moral lines that no responsible technology platform should be permitted to cross. The SAFE KIDS Act draws them clearly in federal law.

We are especially moved by the bill's crisis response requirements. Providers must create and follow a documented protocol for conversations indicating a user is at imminent risk of suicide or self-harm, and must notify parents as soon as feasibly possible when a child linked to a parent account faces such risk [Facts 16, 17]. Crisis response data may not be used for AI training, advertising, or any commercial purpose, and may not be sold or shared except to facilitate the notification itself [Fact 18]. This is a statement of values, not just compliance: a child in crisis is not a data point.

The parental empowerment provisions are among the most thoughtful in the bill. Parents receive a robust settings program with tools to control memory use, data training consent, time limits, and access for children under 13 [Fact 22]. Providers cannot force parents to navigate those settings exclusively through the provider's own AI application [Fact 23]. Providers must actively and regularly promote parental settings through communications designed to reach parents [Fact 24], and must notify parents promptly whenever a child modifies or disables a parent-configured control [Fact 25]. A parent's use of these tools does not waive any legal claim against a provider, except for a claim premised on the notice requirement itself [Fact 26]. We read this as a guarantee that parental engagement is an addition to legal protection, never a substitute for it.

The American Council urges every constituent reading this brief to contact their senator today and ask for a yes vote on SB4855. Children cannot lobby for themselves. Parents deserve the legal backing this bill provides. The faith-informed principle at the center of our work is simple: the strong are responsible for the vulnerable. This bill acts on that principle. Pass it.