SAFE KIDS Act

This bill requires providers of AI chatbots to implement child safety measures, including age estimation, risk assessments, safeguards, parental settings, and independent audits. It also prohibits child targeted advertising and the sale or sharing of children's personal information without parental consent, and gives the Federal Trade Commission enforcement authority.

Sec. 2, Definitions: Defines key terms including "AI chatbot" (an AI system that generates adaptive responses to open-ended input, excluding narrow-purpose systems like customer service or enterprise tools not intended for minors), "child" (under 18), "child safety risk" (foreseeable risk of covered harm), "covered harm" (physical, financial, severe psychological harm, privacy intrusion, or discrimination), "parental setting" (features for parents to manage a child's use), and other terms.

Sec. 3, Determination of user age: Providers must treat all users as children unless they verify the user is 18 or older using age estimation technology. Age signals from such technology are treated as actual age, but providers must use the younger age if signals conflict. Data from age estimation must be minimized, secured, and not retained longer than necessary. Compliance is deemed if a provider meets a substantially similar age assurance framework.

Sec. 4, Additional duties of a provider of an AI chatbot: Providers must conduct risk assessments before launch, before material updates, and annually; implement safeguards for child users (e.g., usage reminders, age-appropriate warnings); not knowingly make available content that promotes eating disorders, unlawful activities, graphic violence, child sexual abuse material, or self-harm; publish a child safety policy; and maintain a crisis-response protocol for conversations indicating risk of covered harm.

Sec. 5, Advertising and information sharing protections for children: Prohibits advertising to child users, child targeted advertising, and selling or sharing children's personal information without verifiable parental consent (with exceptions for judicial process or law enforcement). Providers cannot condition a child's use on consent to data sale or sharing, and cannot use deceptive design patterns that interfere with a child's or parent's choices.

Sec. 6, Rulemaking: The FTC must establish an incident reporting mechanism for third parties to report child safety incidents involving AI chatbots, and create a public online resource listing each AI chatbot's child safety policy.

Sec. 7, AI child safety audit: Providers must annually submit to an independent audit conducted by an auditor meeting FTC standards, testing actual outputs using controlled test accounts. An audit report must be submitted to the FTC and a summary published (with redactions). The FTC may publish aggregated findings.

Sec. 8, Enforcement by the Commission: Violations are treated as unfair or deceptive acts under the FTC Act. Additional civil penalties apply: $1,000 per violation per user for failing to implement safeguards, and $10,000 per violation per user for willful violations involving false information. The FTC may also seek injunctions.

Sec. 9, Relationship to other laws; preemption: The Act does not prevent enforcement of other Federal or State laws that are at least as protective. Compliance does not exempt providers from liability under other laws.

Sec. 10, Severability: If any provision is invalid, the rest of the Act remains in effect.

Sec. 11, Effective date: The Act takes effect 180 days after enactment.