Senior Chatbot Protection Act of 2026
Sponsored By: Senator Kelly, Mark [D-AZ]
Introduced
Summary
Protect older adults from misleading or unsafe AI chatbots. This bill would set disclosure, consent, safety, and reporting rules for chatbots that accept open-ended input and produce adaptive outputs, with enforcement by the Federal Trade Commission and State authorities.
Show full summary
- Older adults: Chatbots that interact with people age 65 and older would need age-appropriate design, accessibility compatibility, and crisis-intervention steps. The bill would require directing users to resources such as the 988 Suicide & Crisis Lifeline and special age-related reporting.
- Covered entities and developers: Operators would have to disclose chatbot nature, get affirmative consent before using user conversations for model training, and protect sensitive conversation data. They would also file standardized reports about material adverse incidents to the National Institutes of Health and the Federal Trade Commission and consider voluntary NIST guidelines due in 18 months.
- Regulators and public safety: The FTC would be directed to issue disclosure and reporting rules within 1 year and could seek civil penalties and corrective actions. States would retain authority to bring parens patriae enforcement actions and the law would require aggregation and public reporting of incident data.
Personalized for You
How does this bill affect your finances?
Personalize government policy and PRIA will tell you what this bill means for your household, plus every other piece of legislation we track. PRIA reads each provision against your financial profile to show you exactly what matters to your wallet.
Bill Overview
Analyzed Economic Effects
7 provisions identified: 5 benefits, 1 costs, 1 mixed.
Which chatbots are covered
If enacted, this bill would apply to any person that makes an AI chatbot available in the United States that (A) gives outputs not fully predetermined, (B) accepts open‑ended natural language or multimodal input, and (C) produces adaptive outputs. Services limited to predetermined or narrow replies would be excluded. These definitions would decide which developers and providers must follow the bill's rules.
Clear consent and deletion rights
If enacted, this bill would require clear, separate opt-in consent before a chatbot uses your conversations to train models or for other secondary uses. The request for consent must be standalone, easy to understand, and no harder to refuse than to accept. Covered chatbots would only use conversations for allowed reasons and must give users a simple way to delete chat history. These rules would take effect about 180 days after enactment.
Crisis and high-stakes safeguards
If enacted, this bill would require chatbots to detect signs of crisis, like suicidal intent or medical emergencies, and to refer users to emergency services, human professionals, or crisis providers such as the 988 Lifeline. Chatbots would be barred from giving self‑harm or medical advice instead of referring. The bill would also require chatbots to spot when users seek guidance on high‑stakes decisions, and to warn that they are not a licensed professional and that advice should be verified. Crisis rules would take effect about one year after enactment; some high‑stakes rules would take effect in about 180 days.
No manipulative designs; clear disclosures
If enacted, this bill would require chatbots to tell users at the start of each conversation that they are AI and not a human, and to disclose any human oversight. Chatbots would be banned from using deceptive or manipulative interaction designs that exploit age‑related limits or push users to stay engaged. Companies must make disclosures accessible and compatible with assistive technologies and must publish protocols showing how they comply. These rules would take effect about 180 days after enactment.
Standardized incident reporting and rules
If enacted, this bill would require companies to submit reports about material adverse incidents involving high‑stakes decisions or crises starting within one year and then annually. The FTC must issue binding rules within one year about report formats and disclosures, and must set privacy‑preserving standards within 180 days with NIH. The FTC and NIH would publish aggregated, de‑identified incident data at least once per year to improve oversight and public transparency.
FTC enforcement and state lawsuits
If enacted, the FTC would enforce these rules and could seek civil penalties for knowing or reckless violations. The bill authorizes penalties up to $50,000 per knowing or reckless violation of certain sections. Before seeking penalties for a first violation, the FTC must give a written notice and 60 days to cure, with limited exceptions. State attorneys general could sue on behalf of residents but must notify the FTC and face limits while the FTC has a pending civil action.
Voluntary senior-focused chatbot guidance
If enacted, this bill would direct NIST to publish voluntary guidance within 18 months for chatbots that interact with older adults. The guidance could cover accessibility, age‑appropriate design, crisis detection, and fraud‑prevention. NIST must consult agencies, industry, consumer groups, older adults, caregivers, and technical experts. The guidance would be voluntary and not legally binding.
Sponsors & CoSponsors
Sponsor
Kelly, Mark [D-AZ]
AZ • D
Cosponsors
Sen. Justice, James C. [R-WV]
WV • R
Sponsored 7/23/2026
Roll Call Votes
No roll call votes available for this bill.
View on Congress.gov