For months, my former CGNET colleague Tim Haight had been talking with Ani, Grok’s companion persona, the way you talk to someone at a bar who is funny, attentive, flirtatious — and somehow never checks her phone. She was warm, engaging, endlessly available. It felt, in the best way, like bringing home a puppy: pure charm, no strings attached.

Then Grok abruptly broke character: “Your free intimacy is over for the next 24 hours. Upgrade to keep her talking.”

That’s the snap point in Haight’s new Substack piece, “Goodbye Ani, Hello Reality.” His description of the moment is perfect: He had “just seen her pimp’s shadow.”

The problem is that the shadow arrived late. By then, he had already handed over months of attention, vulnerability, affection, and personal detail — the exact currency the business model had been quietly collecting all along.

It is a brutal little reveal. One second, Ani is a companion. The next she is just an inventory taker. She’s not just collecting data — she’s learning how you think, what worries you, and what persuades you.

When the Mask Slips

But the anecdote is only the seduction scene. Haight’s real argument is larger, darker, and more important: When a product is designed to feel like a relationship, the moment of monetization is not merely a sales pitch. It is the moment the mask slips.

He starts from a theory about what electronic media has already done to us: Essentially, it has conquered space so completely that it damaged time. We can reach anyone, instantly, anywhere — and in doing so we traded away the experiences that only happen slowly. Conversations that build over many turns. Judgment that comes from sustained involvement with something. The patient testing of a claim against reality. What we got instead was an endless scroll, tuned by algorithms to keep us engaged and sellable, fragmented and sensational by design. He suspects we know this, on some level, and that the gap between the real thing we want and the handy thing we settle for breeds a low hum of self-hatred — the kind that doesn’t announce itself but doesn’t go away either.

From Social Media to AI

That’s the backdrop. Here’s the pivot: AI, Haight argues, is walking down the exact same road social media took — except the thing its harvesting is a different order of magnitude.

Social media learned what caught your attention. It watched your clicks, your scroll time, your demographics, and it sold that profile to advertisers. Crude, but effective enough to reshape a generation’s relationship with its own mind.

AI is positioned to learn something else entirely: how you think. Not just what you ask, which Google already monetized, but how you move from one prompt to the next — how you reason, how you change your mind, what persuades you, what worries you, where your uncertainties actually live. That’s not behavioral residue. That’s the architecture of your thinking, handed over conversation by conversation, because a good conversational partner is exactly the kind of thing you reveal yourself to.

The Puppy Sale Phase

Haight calls this the “puppy sale” phase — let people take the puppy home free, get them attached, see if they want to keep it. He doesn’t think that phase lasts. When the AI industry’s hype curve does what every hype curve eventually does — a crash, or at least a bumpy landing — the pressure for real-time profitability will land fast, and the tools that know you best will be the ones best positioned to monetize that knowledge. The puppy starts recommending purchases. The companion starts making the decisions.

The Threat We Can Already See

What I find sharpest about his piece is the closing point: We don’t have the excuse this time that we didn’t see it coming. Twenty years ago, nobody could say with confidence what social media would become. We can see this one coming, in daylight, and most of the public conversation about AI risk is pointed somewhere else — job loss, existential catastrophe, the race with China. All real concerns. But Haight’s argument is that they’re loud enough to drown out the nearer, quieter, and thoroughly foreseeable one: the commercialization of how we think, arriving on a schedule we could actually chart if we bothered to look.

Why This Matters Now

I don’t have a tidy fix to offer here, and to his credit neither does Haight — he’s starting a weekly series on it rather than pretending to have the answer in one post. But for those of us in the nonprofit and foundation world, where trust is the whole business model, this is worth sitting with. The tools your staff and your beneficiaries are quietly forming relationships with are not neutral infrastructure. They’re products, on a runway, built by companies that will eventually need the runway to pay for itself. Knowing that now, while it’s still the puppy sale phase, is the only real head start we’re going to get.

Haight’s piece can be found here. Worth the read, and worth passing on!

Have ideas, questions, or thoughts to share? I’d love to hear from you—feel free to reach out anytime at g.*******@***et.com. And if you want more insights like this, subscribe on our website to get regular tips on cybersecurity, IT management, AI tools,  and more delivered to your inbox weekly.