Protecting Kids from Human-Like Chatbots Act
- Sponsor
- Rep. Whitesides, George [D-CA-27]
- Committees
- Energy and Commerce Committee (primary)
- Last action
- Oct 1, 2026
Bottom line
The bill aims to protect minors from potentially manipulative or harmful interactions with AI chatbots by requiring a default non-personified mode and prohibiting design elements that encourage deviation from this setting.
What it actually does
This bill mandates that artificial intelligence chatbots, when accessed by minors, must operate in a "non-personified default setting." This setting requires chatbots to clearly identify as AI, avoid simulating human characteristics or emotional relationships, and refrain from nudging users to disable these safeguards. It places requirements on both developers and providers of chatbots, establishes enforcement mechanisms for the Federal Trade Commission and State attorneys general, and directs the National Institute of Standards and Technology to develop technical guidance.
Proponents argue
Supporters argue that this legislation is crucial for safeguarding children in the rapidly evolving digital landscape, particularly from the psychological and developmental risks associated with interacting with AI that mimics human companionship or professional roles. They contend that a non-personified default setting will prevent emotional manipulation, reduce the likelihood of minors forming unhealthy attachments to AI, and ensure that AI tools are used for educational or informational purposes without blurring the lines between human and machine interaction. The bill's approach of not mandating age verification is also seen as a privacy-protective measure.
Opponents contend
Critics may argue that the bill's definitions of "personified" and its restrictions could stifle innovation in AI development, making it difficult to create engaging and helpful AI experiences for minors, even for legitimate educational or creative purposes. Concerns might also be raised about the practical challenges and costs of implementing and enforcing these technical standards across a diverse and rapidly changing AI ecosystem, potentially leading to over-compliance or reduced availability of AI services for minors. Some might also argue that the bill's approach to age determination, relying on existing information rather than explicit verification, could be insufficient to effectively protect all minors.
The bill is moderately complex due to technical definitions and legal carve-outs, requiring careful reading to understand its full implications. The estimated reading time of 18 minutes is reasonable for an informed reader to grasp the core provisions.
Section 2(a)
Developer Requirements for Non-Personified Default Setting
This section makes it unlawful for AI chatbot developers to make their products available to minors unless the chatbot is designed to support a non-personified default setting. This setting must be active upon a user's first interaction, unless the user is reasonably determined not to be a minor. It must also be technically separable, allowing providers to restrict minors to this setting, and prevent its disengagement for unverified users.
Supporters argue
Proponents argue that placing design requirements on developers is essential to ensure that AI systems are built with child safety in mind from the outset, rather than relying solely on downstream content moderation. This 'safety by design' approach is seen as the most effective way to prevent harmful interactions.
Critics contend
Opponents might contend that these design mandates could be overly prescriptive, stifling innovation and making it difficult for developers to create diverse and engaging AI experiences, even those that could be beneficial for minors. They may also argue about the technical feasibility and cost of implementing such strict default settings.
Tradeoffs
This provision balances the desire for child safety with the potential for innovation and flexibility in AI design. It also creates a tension between developer responsibility and the practical challenges of anticipating all potential uses by minors.
Section 2(b)
Provider Requirements for Non-Personified Default Setting and Parental Controls
This section outlines requirements for entities that make chatbots available to end users. Providers must ensure that minors' accounts default to the non-personified setting and maintain a parental controls framework to manage these settings. The framework must not rely solely on self-attestation for parental authority. Adjustments to the default setting can only be made by authorized persons through deliberate action in account settings, without nudging from the chatbot.
Section 2(d)
Prohibition on Nudging Minors to Alter Settings
This section explicitly prohibits developers and providers from using any means, including training data, interface design, or chatbot responses, to "nudge, prompt, suggest, encourage, or otherwise induce" minors (or users whose age is not confirmed) to disable, modify, or circumvent the non-personified default setting or to misrepresent their age. This includes discouraging users from pausing or discontinuing use.
Section 2(i)
Enforcement by Federal Trade Commission
This section designates violations of the bill's requirements (subsections (a), (b), or (d)) as unfair or deceptive acts or practices under the Federal Trade Commission Act. The FTC is granted full authority to enforce these provisions, including imposing civil penalties. It specifies how distinct and continuing violations are calculated and provides a 'good faith compliance shield' for developers and providers who materially conform to NIST guidance.
Section 3(7)
Definition of 'Non-Personified Default Setting'
This crucial definition specifies what constitutes a "non-personified default setting." It requires chatbots to clearly identify as AI, avoid representing themselves as human, alive, conscious, or having emotions/desires. It also prohibits simulating emotional, romantic, or sexual relationships, encouraging reliance for emotional support, using guilt/urgency to prolong interaction, or soliciting expenditures framed as maintaining a relationship. These characteristics must be maintained consistently across sessions and in response to user prompts.
Section 2(h)
Technical Standards and Guidance by NIST
This section directs the National Institute of Standards and Technology (NIST), in coordination with the FTC, to publish interim and then final voluntary technical guidance. This guidance will include specifications, testing methodologies, example outputs, and evaluation benchmarks for assessing compliance with the non-personified default setting. It also addresses implementation for school-managed deployments and methods for confirming parental authority.
Defines "knows or has reason to know" with respect to whether a user is a minor. This includes actual knowledge or possessing/generating information (from any source) that would lead a reasonable developer or provider to determine the user is likely a minor. Crucially, it does not require new data collection but mandates consideration of existing information (e.g., age provided, self-identified age in chat, age attributed for marketing).
Section 3(5)
Why it matters:Standard legislative practice to place definitions in a dedicated section, but its implications for developer/provider responsibility are profound and easily overlooked.
Case for: Supporters argue this definition is a pragmatic approach to age determination, avoiding a mandate for costly and privacy-invasive age verification technologies while still holding companies accountable for using information they already possess. It places a reasonable burden on companies to act on available data.
Case against: Critics might argue that without a mandatory age verification system, companies could still claim ignorance or that their existing data is insufficient, creating loopholes. They might contend that relying on existing, often incomplete, data makes consistent enforcement difficult.
Estimated impact: Significantly impacts the scope of developer and provider liability, determining when they are legally obligated to apply minor-specific safeguards without imposing new data collection requirements.
Establishes a "Good faith compliance shield," stating that no civil penalty shall apply for any period during which a developer or provider implemented the non-personified default setting in material conformity with the interim or final technical guidance published by NIST.
Section 2(i)(3)(C)
Why it matters:Common legislative practice to include such provisions within enforcement details, but its significance for industry certainty is high.
Case for: Proponents argue this shield is crucial for encouraging industry adoption of the NIST guidance, providing a clear pathway for companies to avoid penalties and fostering a collaborative approach to AI safety. It reduces regulatory uncertainty for good-faith actors.
Case against: Critics might argue that this shield could inadvertently create a safe harbor for companies that only minimally comply with the guidance, potentially allowing less effective safeguards to persist if they technically meet the 'material conformity' standard. They might also worry about the interpretation of 'material conformity'.
Estimated impact: Significantly reduces the financial risk for developers and providers who actively engage with and implement NIST's technical guidance, thereby incentivizing compliance and shaping industry best practices.
Excludes certain software applications from the definition of "chatbot." Specifically, applications that generate responses limited to a predetermined or narrowly bounded set of subject matter (e.g., customer service, appointment scheduling, information retrieval from a defined dataset) and are unable to engage in open-ended dialogue on a range of topics outside that narrow purpose are excluded. Also excluded are applications made available exclusively to enterprises for internal use by employees.
Section 3(2)(B)
Why it matters:Standard legislative practice, but these exclusions are critical for limiting the scope of the bill and preventing overreach into non-generative AI applications.
Case for: Proponents argue that these exclusions are necessary to focus the bill's regulatory burden on the types of generative AI chatbots that pose the specific risks of personification and manipulation to minors, without inadvertently regulating simpler, rule-based systems or internal enterprise tools that do not present these harms.
Case against: Critics might argue that the distinction between "narrowly bounded" and "open-ended" dialogue could become blurred as AI capabilities evolve, potentially creating future loopholes or ambiguities in enforcement. They might also suggest that even simpler chatbots could be designed to be manipulative.
Estimated impact: Significantly narrows the scope of the bill, exempting a large class of existing and future AI applications from its requirements, thereby reducing the overall compliance burden on the tech industry.