The U.S. House of Representatives passed the Kids Internet and Digital Safety Act (H.R. 7757) on June 29, 2026, by a bipartisan 267–117 vote. Introduced by House Energy and Commerce Committee Chairman Brett Guthrie (R-KY) alongside Ranking Member Frank Pallone (D-NJ), the legislation bundles 14 distinct child-safety proposals into a single package — and now heads to a Senate where the welcome is noticeably cooler.
What the KIDS Act Actually Does
The package is genuinely broad. It includes a revised version of the Kids Online Safety Act (KOSA), which has circulated in Congress since 2022, with one significant surgical removal: the duty-of-care provision that would have required platforms to minimize a broad, enumerated list of harms to minors. That removal was a real concession. The original duty-of-care language alarmed civil liberties groups and platform lawyers, raising well-founded fears it would incentivize over-removal of legitimate content — from LGBTQ+ resources to addiction recovery forums.
What remains covers substantial ground: age verification obligations using a "knows or should have known" standard tied to user signals and context; expanded COPPA 2.0 protections extending coverage to 13- to 16-year-olds previously excluded from the original 1998 law; restrictions on micro-targeted advertising to users under 17; prohibitions on addictive design features such as infinite scroll and autoplay for minors; and first-of-their-kind federal rules for AI chatbots.
The AI chatbot provisions are the most novel element. Under the KIDS Act, conversational AI tools must make clear and conspicuous disclosures that they are AI systems and not human beings, cannot misrepresent themselves as licensed professionals, must surface crisis resources when users discuss suicide or self-harm, and must prompt breaks after three continuous hours of interaction. These rules address a genuinely new category of harm — AI companions marketed to teenagers that simulate emotional intimacy — and no prior federal statute has addressed this gap.
"The KIDS Act creates strong protections with new rules for design features, default settings, and kids' privacy," Guthrie and Pallone said in a joint statement after passage.
The case for this legislation is not trivial. Algorithmic amplification of self-harm content, predatory behavioral advertising, and AI companions designed to maximize emotional dependency in vulnerable teenagers are documented, real problems. The bipartisan 267–117 margin reflects genuine political will that stretches beyond election-year performance.
The Constitutional Time Bomb
The problem is architectural. The KIDS Act's age-verification mandate applies to any platform that should reasonably know it is being accessed by minors — a standard capacious enough to sweep in most of the consumer internet. To verify user ages, platforms would need to collect government IDs, biometric data, or third-party verified credentials from everyone. That is not a surgical protection for children; it is a de facto national internet identity system.
This design has a direct legal ancestor, and it lost. In Ashcroft v. ACLU (2004), the Supreme Court struck down the Child Online Protection Act (COPA) precisely because its age-verification requirement burdened adult speech in violation of the First Amendment. The Court held that the government must employ the least restrictive means to protect minors — and that parental filtering tools met that standard where mandatory ID gates did not. The KIDS Act's scope is broader than COPA ever was.
The ACLU has urged Congress to vote against the bill, with senior policy counsel Jenna Leventoff stating: "Censorship and invasive age verification measures will not protect children, and the act would threaten free speech for people of all ages and it would put our data at risk." The Foundation for Individual Rights and Expression echoed this, warning that the bill would "eliminate users' ability to speak anonymously" — a right the Supreme Court has protected since McIntyre v. Ohio Elections Commission (1995).
The EFF has specifically flagged that age-estimation systems — the technical backbone of any "should have known" standard — fail disproportionately for people of color, people with disabilities, and trans and nonbinary users, introducing both civil liberties and discrimination concerns into the enforcement mechanism itself.
The AI Chatbot Rules: Useful, but Vague at the Margins
The chatbot provisions are more carefully drafted than the age-verification requirements, but they carry their own ambiguities. A mental health app delivering cognitive behavioral therapy via a conversational AI interface is structurally indistinguishable, under this bill's text, from a social companion app engineered to maximize session length. Requiring therapy bots to interrupt sessions with "I am not a real person" reminders could actively undermine therapeutic relationships that depend on consistency and user comfort.
The prohibition on AI misrepresenting itself as a licensed professional is a different matter — that provision is both targeted and constitutionally defensible. The Senate would do well to tighten the chatbot rules to preserve the good and excise the vague.
The Senate Gauntlet
The bill now sits with the Senate Committee on Commerce, Science, and Transportation, chaired by Sen. Ted Cruz (R-TX), who signaled openness to negotiations but has not set a markup date. The Senate math is significantly harder than the House vote suggests. Sen. Richard Blumenthal (D-CT), a longtime KOSA co-sponsor, said the House version is "dead in the Senate" — a remarkable statement from one of the bill's most ardent supporters, pointing to the deep differences between chambers on duty-of-care obligations, state preemption, and the AI provisions. The August recess and the approaching midterm calendar further compress the available legislative window.
It bears noting that the Senate passed a narrower, standalone KOSA in 2024 with a remarkable 117–2 vote — but that bill's architecture was deliberately more modest than what the House has now assembled into an omnibus. Bundling 14 proposals into one package creates more surface area for objections, more leverage for individual senators, and a harder conference reconciliation if both chambers ever do pass their own versions.
What Proportionate Regulation Would Look Like
The goals animating the KIDS Act are legitimate. The mechanism for its core requirement is not. There is a constitutionally durable path available: a serious update to COPPA — extending its privacy protections meaningfully to users under 17, with real FTC enforcement teeth — paired with aggressive action against data brokers who supply the behavioral profiles that make micro-targeted advertising to children possible in the first place. Restricting the supply of exploitable minor data at the broker layer would disable the advertising business model without requiring every American to hand a government ID to a platform before reading news.
On AI chatbots, the disclosure and crisis-resource requirements are worth preserving with tighter definitions. On age verification, the Senate should heed what the Supreme Court has already decided twice: the least-restrictive-means test is real, and mandatory identity gates fail it.