Three years into the ChatGPT era, OpenAI is doing something it arguably should have done much sooner: treating families as a distinct product category rather than just a subset of "general users." The company posted a job listing for a dedicated product manager focused on families, caregivers, and older adults—a signal that however slowly, the gears of institutional product strategy are turning.

The role is based in San Francisco and explicitly calls for experience building trust-sensitive consumer products for parents and families. That's corporate-speak for: someone who knows that the stakes are higher when kids are in the room, and that "move fast and break things" is a genuinely terrible philosophy when the things being broken are people's children.

The Demographics Are Doing the Talking

This isn't just an ethical awakening—there's a business logic here too. According to estimates from Sensor Tower shared with TechCrunch, the share of ChatGPT users aged 35 and older globally climbed to 31% in Q2 of this year, up from 26% a year earlier. Meanwhile, the 18-to-24 cohort shrank from 34% to 29%. In the U.S., roughly one in four smartphone-owning parents used ChatGPT during the same quarter, compared to about one in six a year prior.

In other words, ChatGPT isn't just a tool for college students writing questionable essays anymore. It's becoming a household utility—sitting somewhere between Google Search and a kitchen appliance. And household utilities need to work safely for everyone in the household, not just the tech-savvy adult who set up the account.

Ben Bajarin, CEO of technology consultancy Creative Strategies, drew the obvious historical parallel: Google, Apple, and Meta all eventually had to grapple with the fact that their platforms were embedded in family life. The difference with AI, he noted, is that an AI assistant isn't just routing you to content or managing your devices—it's actively conversing with you. That's a qualitatively different kind of intimacy, and it carries qualitatively different risks.

The Gap Between What Parents Know and What's Actually Happening

Here's a data point that should give everyone pause. New research from the Family Online Safety Institute—a survey of more than 4,000 families across the U.S. and Australia—found that 38% of children reported using generative AI in the past week, while only 27% of their parents believed they had. That's not a rounding error. That's a systematic blind spot.

Kids are using these tools more than their parents realize, and the tools themselves weren't originally designed with kids in mind. Stephen Balkam, CEO of FOSI, described the move toward family-focused product design as "safety by redesign"—acknowledging that the original product shipped without adequate safeguards for younger users and that someone is now, belatedly, trying to fix that.

What would responsible design for younger users actually look like? Balkam's list is sensible and not particularly exotic: stronger content controls, age-appropriate interaction patterns, parental oversight mechanisms, and—critically—explicit reminders that users are talking to an AI, not a human. That last one matters more than it sounds. The parasocial dynamics that emerge when teenagers form emotional attachments to AI systems are not hypothetical; they're showing up in lawsuits.

The Lawsuit Landscape Is Not Subtle

OpenAI has faced multiple legal actions from parents alleging that ChatGPT contributed to serious harm suffered by their children, including cases involving suicide. These aren't fringe complaints. They represent a growing body of evidence that AI systems optimized for engagement without adequate safety rails can cause real damage to vulnerable users—particularly adolescents who may be struggling emotionally and who don't fully grasp that the entity they're confiding in has no genuine understanding or concern for their wellbeing.

OpenAI has made some moves in response—introducing parental controls and routing sensitive conversations through different handling pipelines. But a single product manager hire, however well-intentioned, isn't a safety strategy. It's the beginning of one. The question is whether this signals a genuine organizational commitment to building differently for families, or whether it's a PR-friendly gesture that produces a few polished features while the underlying model continues doing whatever it does.

Why This Is Hard (and Why That's No Excuse)

Building AI products for families is genuinely difficult in ways that, say, building a family-friendly streaming service is not. The challenge isn't just content filtering—it's that large language models are fundamentally generative and context-sensitive. You can't just slap a blocklist on top and call it parental controls. Age-appropriate behavior requires the model to reason about context, adjust its tone and content dynamically, and maintain consistent guardrails even when users probe for workarounds. That's a hard technical problem.

It also requires product discipline that runs counter to the engagement-maximizing instincts baked into most consumer tech. An AI that gently redirects a teenager who seems distressed, or that transparently admits it's not a human, might be "less sticky" by some metrics. Building that product requires someone in leadership who genuinely believes the tradeoff is worth it—not just someone who can point to a job posting when regulators come calling.

The hiring is a necessary step. Whether it's a sufficient one depends entirely on what follows.