AI queries used as evidence in murder case, highlighting privacy concerns

Prosecutors are presenting a man's queries to artificial intelligence as key evidence in the case of two missing University of South Florida doctoral students. Days before the students were last seen alive in April 2026, the accused allegedly asked ChatGPT about disposing of a body in a garbage bag and how such a crime might be detected. Weeks later, he reportedly searched for the meaning of 'missing endangered adult.'

“

what would happen if a body were placed in a garbage bag and thrown away.

These personal queries stand in stark contrast to the use of AI in military operations, such as the early hours of the US-Israel war on Iran, where AI systems analyzed vast data to accelerate targeting decisions, striking over 1,000 targets in a single day. Both scenarios, despite their vastly different contexts, raise a fundamental question: what are the implications of entrusting consequential information and decisions to machines?

As AI integrates into everyday life, people are casually confiding personal details to these systems. They use AI to process relationships, rehearse conversations, unpack trauma, and make personal decisions, describing workplace conflicts, financial struggles, health anxieties, and family tensions. This behavior is driven by AI's non-judgmental, patient, and often empathetic responses, which can be particularly helpful for individuals experiencing loneliness or emotional isolation.

“

how such a crime might be detected.

However, the perceived safety and usefulness of AI should not be mistaken for genuine privacy. The more context a user provides to an AI system, the better its performance, but this also means potentially divulging sensitive information like names, locations, medical details, and confessions. This data can become part of a record held on systems beyond the user's control. Most users do not read privacy policies, and the illusion of privacy is often maintained by convenience rather than robust protection.

“

what does it mean to hand consequential information, and sometimes consequential decisions, to machines?

The risks associated with this data sharing are not hypothetical; they are already manifesting. Incidents like Samsung engineers reportedly entering proprietary code into ChatGPT while debugging software have led to data leaving internal systems and entering public AI environments. Similar breaches have occurred across industries as professionals continue to input confidential documents into AI tools.

“

AI may feel private, but personal and private are not the same thing.