We’re freaked out by how tailored our advertisements on Instagram are, or how much Pinterest seems to anticipate our interior design taste — and yet we continue to spoon-feed AI chatbots heaping piles of intensely personal information.
AI companies are now in possession of heaps of personal information that social media giants could only dream of. We aren’t just setting ourselves up for hyper-targeted ads. We’re also making ourselves vulnerable to hacks and even blackmail.
ChatGPT alone receives 3 billion messages from users daily. Every time we ask a bot for advice, vent to it, or emote at it, we’re contributing to a treasure trove of hyper-personal information, even though we should know better.
Considerable proportions of Americans who use AI say they turn to it for frighteningly personal tasks: 20% for medical advice, 20% for diet information, 10% for emotional support, and 4% for companionship. Those are just self-reported numbers; data from OpenAI itself is even more revealing.
Their internal analysis of 1.1 million messages sent in 2025 found that 73% of chats were not work related. Eight-point-five percent of messages were seeking personalized how-to advice, 5.7% health, fitness, beauty, or self-care, 5.3% were considered “self-expression,” and 1.9% personal reflection or talking about relationships.
Potential for rogue bots deciding they might exploit the data of users isn’t remote. During a 2025 test, one Anthropic model showed a real knack for blackmail. Facing the fictional threat of being disconnected, the model went after a fake executive and threatened to expose email evidence of his marital affair unless he intervened to prevent the shutdown.
We’re stuck in a fear-versus-behavior paradox. Half of US adults use AI as of this year, and a quarter use it every day. And yet 7 in 10 Americans admit they believe using AI will make their data less secure.
Perhaps we’re willingly baring it all to Big Tech because we’ve become so dependent on their products. But is the risk really worth the reward? What happens if your AI bot of choice starts harvesting data for its makers to sell? What if a major AI company gets hacked? Or a rogue bot starts blackmailing users?
What if insurance companies start mining your chatbot conversations about medical symptoms or your smoking habit? Or a loan provider decides they’d like to know what you say about your personal finances when only a robot is around?
And what if your personal chat logs end up being used as evidence in court? That possibility isn’t even hypothetical. Law enforcement, provided they have a warrant, can access your AI chat logs and conversations as readily as your Google searches.
Already, AI chat logs have been used as evidence in murder investigations, like in the case of former NFL player Darron Lee who allegedly asked a bot to help cover up his girlfriend’s death. Criminals are learning the hard way that AI isn’t a black box, including one Missouri student who allegedly asked AI “how f—ed am I bro” while confessing in detail to vandalizing cars.
Regular users of chatbots seem to be literally or willfully ignorant of the privacy concerns, but Silicon Valley leaders certainly aren’t.
Palantir CEO Alex Karp demanded in a July CNBC interview that big AI firms answer the pretty basic questions: “Who owns the data? Where is it cached? Are the prompts secure?”
We should all demand answers to these questions before typing away into a robot’s database — and yet we keep volunteering our personal information anyways, because it’s convenient.
But the most concerning dimension of our privacy nightmare is how much information AI companies have about their users.
For instance, Anthropic said in their September 2026 threat intelligence report that they identified and banned users who were using their AI for the purpose of fraud, scams, cyberattacks, weapons development, and influence operations.
While, of course, we should all be glad that those users were stopped, the fact that they’ve been singled out is proof that your conversations with AI aren’t simply going into a digital void. In fact, they’re going into a very valuable pile being accessed by law enforcement and leveraged by companies for profit.
OpenAI is already actively developing an advertising arm of their business. On their website, they excitedly tell potential advertisers that they’ll be able to “reach people in moments that matter,” because “people come to ChatGPT… [to] make decisions.”
“Go beyond keywords with richer context signals,” the company pitched on their landing page for advertisers. “In ChatGPT, people share richer context, enabling advertising that is more relevant, personalized, and useful.”
Of course, OpenAI has stressed that they “keep your conversations with ChatGPT private from advertisers” and “never sell your data to advertisers.” The company referred The Post to public statements assuring users that they can opt-out of personalized ads.
But AI companies’ pivot into the advertising business represents a new frontier in surveillance capitalism. We’re moving past the era when social media companies predicted what you’d like based on your clicks. Now, we’re telling Big Tech our preferences, our fears, our dreams, our desires and our secrets.
AI has become our stylist, our friend, our confidant, our therapist, even our doctor — and we’re avoiding grappling with what that really means for our privacy, let alone our humanity.
