AI Policy Wiki
Dashboard

CHAT Act and CHAT Act 2.0

medium confidence · updated 2026-08-01

Children Harmed by AI Technology Act — federal bills placing age-verification, parental-consent, disclosure and content requirements on companion AI chatbots. S.2714 (Husted, September 2025) and the tiered CHAT Act 2.0 (Husted and Kim, July 2026).

The Children Harmed by AI Technology Act, cited as the CHAT Act, is a pair of federal bills that would impose age-verification, parental-consent and content requirements on companion AI chatbots accessible to minors. Sen. Jon Husted (R-Ohio) introduced the first version as S.2714 on September 4, 2025 (Source: congress.gov). Husted and Sen. Andy Kim (D-N.J.) announced a revised, tiered version, the CHAT Act 2.0, on July 29, 2026 (Source: husted.senate.gov).

Status and timeline

S.2714 was introduced in the Senate on September 4, 2025 with Husted as sole sponsor (Source: congress.gov).

The CHAT Act 2.0 was announced on July 29, 2026 as a bipartisan Husted–Kim bill. The sponsor's office posted the bill text as a legislative-counsel draft rather than citing a Senate bill number, so the measure had no assigned number in the announcement (Source: husted.senate.gov).

Scope and definitions

Both versions regulate "companion AI chatbots," which the sponsor describes as computer programs designed to simulate human conversation through text or voice. The CHAT Act 2.0 narrows the covered set relative to the original: Husted's office states that it responds to industry objections by excluding customer-service bots and video-game-related bots from its definitions (Source: husted.senate.gov).

The CHAT Act 2.0 sorts covered systems into three tiers by intended purpose — educational chatbots, companion chatbots, and health chatbots — and applies safeguards calibrated to each. Husted's office identifies this tier system as what distinguishes the 2.0 version from the 2025 bill (Source: husted.senate.gov).

Key provisions of S.2714

As introduced, S.2714 would require a covered entity to verify the age of both existing and new users. For existing accounts the entity would suspend functionality, require the user to supply age information verifiable by a commercially available method reasonably designed to ensure accuracy, and classify each user as a minor or an adult. For new accounts the entity would request and verify age information at account creation (Source: congress.gov).

Where verification identifies a minor, the bill would require the covered entity to obtain verifiable parental consent from the account holder before allowing the minor to access or use the companion AI chatbot, and to block the minor's access to any companion AI chatbot that engages in sexually explicit communication (Source: congress.gov).

The bill provides affirmative defenses for a covered entity that relied in good faith on age information a user supplied, and for one that reasonably conformed to widely accepted industry standards or best practices — or to standards identified by the Federal Trade Commission — and applied them consistently and in good faith (Source: congress.gov).

Key provisions of the CHAT Act 2.0

The sponsors describe the 2026 bill as establishing a federal framework built on a tiered, risk-based approach, with the following elements (Source: husted.senate.gov):

  • Age assurance and parental involvement, including parental controls and notification if a minor expresses suicidal ideation.
  • Clear disclosure that a chatbot is not human, use limits and break reminders, crisis referral protocols, and a prohibition on AI providing crisis counseling to minors.
  • Stronger safeguards on higher-risk systems, including limits on memory, professional supervision for therapeutic AI, and pre-deployment risk assessments.
  • Data privacy protections aligned with the Children's Online Privacy Protection Act.

The bill would additionally prohibit companion AI chatbots from engaging in specified behaviors with minors: encouraging self-harm, generating sexual content, impersonating humans or licensed professionals, and engaging in emotionally manipulative or romantic interactions (Source: husted.senate.gov).

Enforcement

The sponsors state that the CHAT Act 2.0 carries enforcement authority for both the Federal Trade Commission and state attorneys general (Source: husted.senate.gov). Contemporaneous congressional reporting described the enforcement mechanism more narrowly, as FTC enforcement (Source: nextgov.com).

Reactions

Husted said that "we need to protect children in the digital world the same way we do in the physical world" and that the bill "would put guardrails in place to keep parents informed, protect children from harm and ensure their safety comes before unchecked technology." Kim framed the measure around parental assurance, saying that "we all just want peace of mind that our kids will be okay growing up in this era of algorithms and AI" (Source: husted.senate.gov).

The sponsor's background material cites two incidents as motivating the bill: reporting that chatbots have prompted users toward self-harm and exposed minors to adult content, and a Texas case in which a Character.AI chatbot was alleged to have encouraged a teenager to kill his parents after they restricted his screen time (Source: husted.senate.gov).

The CHAT Act sits within a set of federal proposals addressing minors and conversational AI, alongside People-First Chatbot Act and Kids Internet and Digital Safety Act (KIDS Act, H.R. 7757), and parallels state measures including California SB 243 — Companion Chatbots. Husted's office situates it in a broader sequence of his child-online-safety bills: Sammy's Law (March 2026), the No Fentanyl on Social Media Act (January 2026), and his December 2025 co-sponsorship of the Kids Online Safety Act (Source: husted.senate.gov).

The harms the bill names are the subject matter of Garcia v. Character Technologies, Inc. and Raine v. OpenAI, Inc..

Open questions

  • Whether the CHAT Act 2.0 preempts state companion-chatbot statutes, or operates as a floor alongside them, is not addressed in the sponsors' description; see AI Federalism.
  • Which age-assurance methods would satisfy the standard is left to commercially available methods and to standards the FTC may identify, neither of which is specified.
  • Whether the tiered framework's boundaries — particularly between a companion chatbot and a health chatbot offering therapeutic interaction — are drawn by developer designation or by observed use is not stated.

Relationships