On April 13, 2026, days before two doctoral students at the University of South Florida were last seen alive, the man suspected of killing them allegedly asked ChatGPT what would happen if they put their bodies in a garbage bag and threw them away. He later asked how such crimes were discovered. A few weeks later, he looked up the meaning of “missing endangered adult.” Prosecutors are now pointing to these interrogations as part of their examination of evidence.
Around the same time, artificial intelligence was being used on a completely different scale. In the early days of the US-Israel war against Iran, more than 1,000 targets were attacked per day. Military officials said AI systems were being used to analyze vast amounts of data and speed up targeting decisions.
These are very different situations. But they raise the same fundamental question: What does it mean to hand over important information, and in some cases important decisions, to a machine?
For most people, the risks are not that dramatic. But the same logic enters everyday life. What does it mean to confide in an AI? What happens to the information we share?
We communicate everything to machines quite casually.
Around the world, people are using AI to manage relationships, rehearse difficult conversations, unravel trauma, and make personal decisions. They detail conflicts at work, financial struggles, health concerns, and family tensions. They do this at scale, often without hesitation, and often without considering where that information is going.
This behavior is not unreasonable. AI can give you peace of mind. Don’t interrupt, criticize, or gossip. They respond immediately, patiently, and often with an empathetic tone that many people find lacking elsewhere. That resilience can be really helpful for people who are working through loneliness, burnout, or emotional isolation.
However, usefulness should not be mistaken for safety. AI may feel private, but personal and private are not the same thing.
The central tension is simple. The more context you give an AI system, the better it tends to perform. But the more you give, the more you may end up giving. Names, documents, locations, medical details, legal concerns, and intimate confessions can all become part of records held on systems you don’t control.
Most users don’t read privacy policies. Few people still understand how data is stored, reviewed, and potentially reused. The illusion of privacy is maintained not by protection but by convenience.
Risk is not hypothetical. They are already visible at both organizational and individual levels.
In 2023, a Samsung engineer reportedly entered proprietary code into ChatGPT while debugging software. The data leaked from internal systems and entered the public AI environment, prompting Samsung to restrict such use. Similar incidents continue across the industry as experts continue to paste sensitive documents into AI tools without fully considering the consequences.
On a personal level, the risks are just as real. When you share identifiable information, legal documents, medical records, or detailed personal stories, you create a digital footprint that doesn’t easily go away. Many AI platforms preserve conversations by default. Some use them to improve their models. Some allow human reviews. Policies vary, but the more important point remains the same. The thing is, most users don’t fully understand what happens after they press send.
The USF incident makes that reality clear. Even after the conversation ended, the suspect’s interrogation did not go away. These became part of the record that could be captured, analyzed, and used.
This is not an argument against AI. It is a cautionary argument.
Some risks are less obvious, but just as important. AI is becoming emotionally convenient. You can listen without getting tired. You can test it without any resistance. It reflects your words and tone in a way that makes you feel attentive and caring.
However, there is a difference between being heard and being restrained.
AI is not responsible. You won’t face any consequences. We will not challenge you unless asked to do so. Over time, that ease can give way to more demanding forms of human interaction. The danger is not necessarily that the AI will directly harm you. That means it can be an easier alternative to relationships that require effort, commitment, and discomfort.
We are entering a phase where the real skill is not just knowing how to use AI, but knowing how to limit it.
It requires a change in behavior. Before you share anything, it’s worth asking a simple question. Should I say this in a crowded cafe, knowing that a stranger might overhear me, or should I say this in a pair of tongs on the side of the road? If not, then it probably doesn’t belong in a chatbot either.
Anonymity should be the default. Specificity must be intentional. Documents must be handled with care. At a minimum, users should understand the basic data policies of the tools they utilize.
These aren’t just technical tweaks. They are cultural. It’s not just privacy that’s at stake, but judgment.
Most of us will never face a situation where a suspect’s AI search history becomes evidence in a court of law, or where automated decision-making at the national level raises fundamental questions about human responsibility. But the underlying logic is the same at all scales. AI systems can retain what we share, and that information can travel far beyond the given moment.
We are living through an early, largely unregulated stage of AI in our daily lives. Technology is advancing faster than the surrounding standards. In that gap, the burden of attention falls heavily on the user.
The new digital literacy of this era may not be about learning how to use AI. Maybe I’m learning what not to give. Knowing where to draw the line is becoming wisdom.
Tricia Nashtaran Futurist and human-centered design expert.
