AI Moved Into Your House and Nobody Asked

The Family Plan Nobody Requested

OpenAI is hiring a product manager to build ChatGPT experiences for families, caregivers, and older adults. Let that sink in for a second. Three years after ChatGPT launched as a productivity tool for knowledge workers and curious nerds, the company is now thinking about households. Not users. Households.

The demographic shift is real, according to Sensor Tower data shared with TechCrunch: the share of ChatGPT users aged 35 and older globally rose to 31% in Q2, up from 26% a year earlier. In the U.S., nearly one in four smartphone-using parents used ChatGPT during the quarter. So OpenAI isn't chasing a hypothetical market. It's chasing the market that already showed up uninvited.

Ben Bajarin of Creative Strategies compared this to the path Google, Apple, and Meta followed as their platforms became embedded in daily life. Fair enough. But here's what bugs me: those companies had years of iteration before they started building "family" features, and they still got it wrong constantly (remember YouTube Kids?). OpenAI is attempting this with a product that hallucinates, that can be manipulated by a clever 12-year-old, and that has no real track record of age-appropriate guardrails. Stephen Balkam of the Family Online Safety Institute called it "safety by redesign," which is a polite way of saying the original design didn't account for kids at all.

And here's the detail that should worry every parent paying attention: new research from the same institute found that while 27% of U.S. parents said their child had used generative AI in the past week, 38% of children said they had. The gap between what parents think is happening and what's actually happening is already wide. Building a family product doesn't close that gap. It might widen it, by making parents feel like the problem is solved.

Meta's Predictable Stumble

OK so Meta rolled out a feature earlier this week that let users generate AI images by @-mentioning public Instagram accounts they wanted to reference. The feature didn't notify the person whose photos were being used. It took approximately zero seconds for everyone to realize this was going to be abused to generate, among other things, naked images of people without their consent. Meta pulled it on Friday.

The company's statement is almost comically corporate: "Our intent was to provide a useful creative tool and to give people control over whether their public content could be referenced in this way. We've heard the feedback that this feature missed the mark, so it's no longer available." Missed the mark. Like they were aiming for a dartboard and hit the wall. As TechCrunch noted, talent agencies including CAA were already raising alarms. The fact that this feature shipped at all tells you something about the internal review process at Meta Superintelligence Labs (and yes, that's really what they call it).

What's interesting isn't that Meta pulled the feature. Companies pull bad features all the time. What's interesting is the pattern: ship fast, absorb backlash, retreat, reframe it as "listening to feedback." This is the third or fourth time Meta has done exactly this dance with AI features on Instagram in the past year. At some point you have to wonder whether the backlash itself is part of the product strategy, a way to test boundaries and see what sticks. Maybe I'm being cynical. But the alternative is that one of the largest technology companies on earth genuinely didn't foresee that letting strangers remix your photos with AI might go badly, and I'm not sure which explanation is worse.

Apple Draws Blood

Apple filed a lawsuit on Friday against OpenAI alleging trade secret theft and breach of contract. This is not a vague IP dispute. It's specific and ugly. Apple accuses OpenAI's Chief Hardware Officer Tang Tan, who spent 24 years at Apple, of using Apple's confidential project code names during recruiting, asking job candidates to bring Apple hardware components to interviews, and coaching departing employees on how to evade Apple's security procedures.

Another former Apple employee, Chang Liu, allegedly failed to return a company laptop after joining OpenAI in 2026 and used it to download confidential technical documents. He's also accused of sharing Apple's confidential information with other Apple employees who were applying to OpenAI. Apple says it sent a letter to OpenAI in February raising these concerns and received no response. No response! That's either breathtaking arrogance or spectacularly bad legal counsel.

The timing matters enormously. OpenAI is rumored to be developing its first hardware product, possibly a smartphone that relies on AI agents instead of apps, according to analyst Ming-Chi Kuo. Last year OpenAI acquired Jony Ive's device startup io for $6.5 billion. So this isn't a lawsuit about the past. It's Apple trying to kneecap a future competitor before it gets off the ground. If the allegations hold up (and the specificity of the claims, code names, laptops, coaching on security evasion, suggests Apple has receipts), this could get genuinely messy for OpenAI's hardware ambitions.

The Infrastructure Underneath

While everyone was watching the drama, the actual plumbing of the AI industry had its biggest week in years. SK Hynix raised $26.5 billion in its U.S. market debut on Friday, the largest-ever IPO by a non-American company, topping Alibaba's $25 billion in 2014. The stock opened 14% above its IPO price. Demand was reportedly seven times the available shares.

SK Hynix makes high-bandwidth memory, a key component of Nvidia's AI GPUs. The fact that a memory chip company from South Korea can generate this kind of Wall Street frenzy tells you where the real bottleneck in AI is right now. It's not models. It's not software. It's silicon. U.S. Commerce Secretary Howard Lutnick used the moment to push SK Hynix and Samsung to build factories in the U.S., which is the kind of gentle suggestion that comes with implied tariff threats.

Meanwhile, Hugging Face CEO Clem Delangue made the case on TechCrunch's Equity podcast that open-source AI is booming precisely because companies are tired of renting their intelligence from frontier API providers. The pattern he described is simple: companies start on OpenAI or Anthropic's APIs, costs balloon as they scale, and they migrate to open models. Roughly half the Fortune 500 now uses Hugging Face. Delangue also flagged something worth watching: Chinese labs are producing the majority of open models being downloaded in the U.S., and he thinks that's a problem worth fixing rather than a reason to distrust open source itself.

There's a thread connecting all of this week's stories that nobody's quite saying out loud. OpenAI wants to be in your living room. Meta wants to be in your camera roll. Apple wants to make sure nobody else gets there first. SK Hynix is selling the picks and shovels. And Hugging Face is betting that eventually, companies (and maybe families too) will want to own the AI rather than borrow it. The question isn't whether AI becomes domestic infrastructure. It already is. The question is whether anyone building it has thought seriously about what happens when the thing you invited into your house starts acting like it owns the place.