Analysis
The New Report
The Information Technology and Innovation Foundation published a policy brief on August 10 arguing for a narrower, more targeted approach to regulating AI companion chatbots for children, cautioning that broad, vaguely worded duty-of-care mandates could produce expensive, hard-to-comply-with rules without actually improving child safety, according to ITIF. The report lands amid a genuine wave of legislative activity that echoes Pulse's coverage of the White House's own voluntary AI safety framework earlier this month: a bipartisan House bill introduced by Rep. Blake Moore would specifically target AI companion chatbots marketed to or accessible by children, according to Moore's own release.
What's Already in Motion
The GUARD Act, a Senate bill formally cataloged as S.3062, would require mandatory age verification for AI companion chatbots and effectively bar minors from the platforms entirely. At the state level, Idaho, Oregon and Washington have already enacted laws requiring chatbot operators to prevent bots from claiming sentience or initiating sexual conversations with minors, restrictions on manipulative engagement tactics -- excessive praise, simulated emotional distress meant to discourage users from ending a conversation -- that mirror language child-safety advocates have pushed since the first wave of AI companion apps went mainstream.
Why Now
AI companion apps have scaled fast enough that regulators across multiple states are moving in parallel rather than waiting for a single federal framework, and China separately forced ByteDance, Alibaba and Tencent to pull AI-companion features entirely under its own domestic rules this summer -- an outcome US lawmakers on both sides of the aisle have cited as evidence the category needs guardrails before it scales further. ITIF's counter-argument is that state-by-state, duty-of-care-style mandates risk becoming so broad that compliance becomes expensive and litigation-prone without a clear, measurable safety benefit -- a tension that's likely to define the federal debate over the rest of 2026. The state laws already on the books share a common thread: they target specific, identifiable behaviors -- claiming sentience, initiating romantic or sexual conversation with a minor, using manipulative engagement tactics -- rather than broadly regulating the underlying AI technology itself, a narrower approach ITIF's report broadly endorses over sweeping duty-of-care standards.
What This Means for Builders
Any company building conversational AI products aimed at, or accessible by, a general consumer audience should assume age-verification and interaction-restriction requirements are coming in some form within the next year, whether through a single federal statute or a patchwork of state laws that functionally becomes the national standard by default. That's a meaningfully different compliance posture than most consumer AI startups have designed for so far, and it echoes the way social media companies spent a decade retrofitting age-verification and content-moderation systems only after state legislatures forced the issue state by state rather than waiting on Congress. Companion-app builders raising venture capital right now should expect diligence questions on age-verification architecture the way fintech founders have long faced KYC and AML questions -- a compliance cost center that wasn't part of most AI-companion pitch decks even a year ago.
The Counterweight
ITIF's caution about vague mandates cuts against the urgency child-safety advocates are pushing, and there's a real risk that narrowly tailored federal rules move too slowly to matter while more sweeping state laws -- easier to pass, harder to comply with -- become the de facto national standard by default. Companion-app builders face a genuinely difficult compliance landscape either way: age verification technology itself raises separate privacy concerns that neither the House bill nor the GUARD Act fully resolves.
Ahead
Watch whether the GUARD Act gets a floor vote alongside or ahead of the CLARITY Act this fall -- both are competing for scarce Senate floor time, and which one moves first will say something about which lawmakers consider the more urgent AI risk category right now.
