Who Are We Really Protecting? The Double Standard Around AI Companions

Here's a question I keep coming back to: if we're genuinely alarmed about what AI is doing to teenagers, why are we so much more comfortable promoting those same tools to elderly people, including those with cognitive decline?

That inconsistency deserves a closer look. Not because the answer is simple, but because nobody seems to be asking the question out loud.

When it comes to younger generations, the concern is clear and well-documented. Researchers, parents, and policymakers are raising real alarms about dependency, emotional manipulation, and the replacement of human connection with simulated intimacy. Those concerns are grounded in evidence. They make sense.

So what changes when the person holding the device is 78, living alone, and showing early signs of memory loss?

Part of the answer is caregiver exhaustion. And I say that with full compassion, not criticism. If you are caring for a loved one from a distance, or managing their care while also raising children and holding down a job, the promise of a tool that will keep them "safe and engaged" lands differently than it does for a parent worried about their teenager's screen time. The desperation is real. The need is real.

But desperation is also exactly the condition that makes people vulnerable to marketing language that sounds like care.

Here's what that marketing language tends not to mention. The loved one is still alone. They're having some of the most intimate conversations of their day with a device. And those conversations are being stored, by a company, under terms of service that most families never read. We're talking about vulnerable people sharing their fears, their memories, their confusion. The long-term implications of that data are almost never part of the conversation.

There's also an assumption embedded in our collective comfort with this that's worth naming directly. We seem to believe it matters more for a developing brain to engage with the real world than for a declining one. That once cognition is deteriorating, any connection is better than none. That the bar for what counts as meaningful engagement can simply be lowered.

That's a values statement. And it deserves more scrutiny than it's getting.

This connects to something writer Charlie Warzel explored recently in The Atlantic: the growing sense that we're losing control of our own information environment. We believe we're researching. We think we're forming independent opinions. But what we're actually consuming has already been shaped for us, by algorithms and AI-generated content and platforms designed to maximize engagement, not understanding. That includes the information families use when evaluating care tech products for their loved ones. The "research" we're doing is rarely as independent as we believe.

So what do I want you to take from all of this?

Not fear. Not a verdict against AI companions. There are genuine needs here: real loneliness, real gaps in care. And technology, when used thoughtfully, can play a meaningful role in addressing them.

But meaningful doesn't mean unexamined. The thoughtfulness we bring to conversations about teenagers and technology — the questions about consent, dependency, data, and the difference between connection and the simulation of connection — should apply here too.

These are conversations the caregiving community deserves to be having openly. Not just between Caregivers making difficult decisions in difficult moments, but between Caregivers and the companies building the tools they rely on. Ask the hard questions. Expect real answers. And if the companies marketing these products aren't ready to have that conversation, that tells you something too.

You deserve the full picture, we all do. Full stop.