The GUARD Act, Explained: The Federal Bill That Would Age-Verify Every AI Companion

The GUARD Act would prohibit anyone under 18 from using AI companions, require operators to verify user age, and require chatbots to disclose at the start of every conversation and at intervals that they are not human and not a licensed professional. It cleared the Senate Judiciary Committee 22-0 on April 30, 2026. It is not law.

That gap between committee vote and law is where most coverage of this bill goes wrong, so it is worth holding onto while reading the rest.

What the bill actually says

The GUARD Act, formally the Guidelines for User Age-verification and Responsible Dialogue Act, was introduced in the Senate in October 2025 by a bipartisan group including Josh Hawley, Richard Blumenthal, Katie Britt, Mark Warner and Chris Murphy, with a House companion bill from Blake Moore and Valerie Foushee.

Four things sit at its center.

Minors cannot access AI companions. If an age verification process determines a user is under 18, the covered entity must block that user from any AI companion it makes available.

Operators must run age verification. The bill specifies the requirement and the outcome rather than prescribing one technical method.

Chatbots cannot claim to be licensed professionals. No representing itself as a therapist, physician, lawyer or financial advisor, directly or indirectly.

Chatbots must disclose their nature repeatedly. At the start of each conversation and at reasonably regular intervals, a chatbot must state that it does not provide medical, legal, financial or psychological services and that users should consult a licensed professional.

Penalties reported at up to $250,000 per violation sit behind it, with criminal provisions aimed at operators who knowingly ship chatbots that engage minors sexually or encourage self-harm.

The definition is the load-bearing part

Legislation like this lives or dies on scope, and the committee amendment that mattered most narrowed the age-verification requirement to companion chatbots specifically rather than general-purpose models.

The bill defines an AI companion as a chatbot that simulates a sustained interpersonal relationship or emotional interaction, by either exhibiting persistent responses suggesting affection or attachment, or presenting at least one persistent identity or character and holding itself out as a sentient being, fictional character or social entity.

Read that second clause carefully, because it is broader than the AI girlfriend category the headlines describe. A persistent fictional character is a companion under this definition. That covers roleplay platforms, character libraries, and a meaningful slice of products that market themselves as creative tools rather than relationships. It plausibly does not cover a general assistant with no persistent persona, which is what the amendment was for.

Why this bill and not the last dozen

The pressure did not come from nowhere. The FTC opened an inquiry into seven companies operating companion-style chatbots in September 2025. Multiple wrongful-death suits proceeded through 2025, and by January 2026 Google and Character.AI had agreed to settle the leading cases.

Meanwhile the category kept growing. Companion apps were downloaded around 220 million times globally as of mid 2025, and survey work in 2026 found roughly one in ten adults using AI chatbots for emotional support and one in twenty-five for companionship. MIT Technology Review named AI companions one of its breakthrough technologies of 2026.

A large, fast-growing consumer category with active litigation and an open federal inquiry is the standard recipe for legislation actually moving, which is why a unanimous 22-0 committee vote happened here and not on the previous rounds of chatbot bills.

The states already did some of this

Federal action would land on top of a state layer that is already live rather than on empty ground.

Illinois passed the first state ban on AI delivering therapy, effective August 2025, with penalties reaching $10,000 per violation. Nevada barred AI from providing mental or behavioral health care from July 2025. Utah regulated rather than banned mental health chatbots, requiring AI disclosure, blocking the sale of health data and barring in-chatbot advertising. Texas brought chatbot disclosure duties in from January 2026. California's SB 243 addressed companion chatbots directly, and we broke that one down separately in what the Companion Chatbots Act actually requires.

Idaho, Oregon and Washington have since enacted rules requiring operators to prevent chatbots from claiming sentience or initiating sexual conversations with minors.

No state currently mandates upfront ID verification for general AI access. The enacted laws mostly trigger on actual knowledge of a user being a minor, through self-declaration or detected signals, which is a considerably softer standard than what GUARD proposes.

What this would mean for adult users

The realistic answer is friction, and the source of the friction is worth naming precisely: blocking minors requires identifying them, and identifying them means checking everyone.

The category has an unresolved rating problem that makes the case for legislators. A survey of more than 4,000 companion apps found the majority carried store ratings suitable for minors. Character.AI itself is rated 17+ on the App Store and Teen on Google Play, the same company submitting the same product with two different declared audiences.

For adults, the practical outcomes of a passed GUARD Act would likely be some combination of ID or document checks at signup, more platforms defaulting to web rather than app store distribution, and smaller operators exiting the US market rather than building compliance. That last one is not speculation. It is what happened to several platforms in adjacent categories under state age-verification laws, and we tracked the international version in where AI companions are actually illegal right now.

The honest odds

A unanimous committee vote is a real signal and not a guarantee. Plenty of bills clear committee 22-0 and never see a floor vote in a crowded session. Age verification regimes also draw First Amendment challenges over burdening adult access to lawful content, though the Supreme Court's recent application of a lower scrutiny standard to a Texas adult-site law shifted that terrain in favor of the laws.

If you use these platforms, the sane posture is neither panic nor dismissal. Know that the disclosure requirements are the cheapest part for operators and the most likely to appear regardless, that verification is the expensive part and the fight, and that your own data exposure grows the moment a platform starts collecting identity documents. Our AI companion privacy guide covers what that actually means for what you have already sent.

For the broader question of whether Western bans are plausible at all, we took that apart in could the West ban AI companions.