If ChatGPT Were Human, Would It Face Criminal Charges?
This legal dilemma might sound like science fiction, but it is now at the center of a real case. In April 2025, Phoenix Ikner, a Florida State University student, reportedly exchanged messages with ChatGPT about which weapons and ammunition to use, as well as when and where to carry out an attack. He then shot multiple people on campus, resulting in two deaths and six injuries.
Florida state prosecutor James Uthmeier said his office’s investigation pointed to ChatGPT being involved in planning details of the attack. This led to an ongoing criminal investigation, with Uthmeier stating he could consider indictments against OpenAI, which developed ChatGPT, or its employees.
“If that thing on the other side of the screen (ChatGPT) were a person, we’d charge it with homicide,” Uthmeier remarked a few days after the tragedy.
Are AI Companies Criminally Liable?
While US law does allow for criminal charges against companies and their employees, such cases are relatively rare. Purdue Pharma, for example, was fined $8.3 billion for its role in the opioid crisis. Volkswagen faced criminal sanctions for emissions cheating, while Pfizer and Exxon were prosecuted in the Bextra drug and Exxon Valdez oil spill cases, respectively.
Legal experts say the most likely charges against AI companies would be negligence or recklessness (willful disregard for known risks or safety obligations). These are generally classified as misdemeanors, with less severe penalties than felony offenses.
“Because this is new legal territory, a convincing prosecution would likely require internal documents showing the company was aware of the risks of deadly misuse, but didn’t take them seriously enough,” said Matthew Tokson, a law professor at the University of Utah. “It would probably need to be overwhelming to persuade a jury or a judge.”
OpenAI Maintains Innocence
OpenAI stated, regarding the Ikner case, that
“ChatGPT is not responsible for this horrific crime.”
The company says it is
“constantly working to strengthen our safeguards to detect dangerous intentions”
and
“respond appropriately if there is a risk to anyone’s safety.”
Brandon Garrett, a Duke University law professor, emphasized,
“the burden of proof is much heavier in criminal law,”
and that doubt generally benefits the defendant. Garrett added that a tech employee could only be implicated indirectly, if at all.
“It’s a product (ChatGPT) that may have encouraged the crime, not direct human intervention,”
Tokson said.
“That’s what makes it so interesting.”
Civil Suits: An Easier Path?
According to Tokson, civil actions
“would be more promising,”
since plaintiffs can rely on broader concepts of liability. Multiple lawsuits have already been filed against US AI platforms, often in connection with suicides, but none have resulted in a judgment so far. In December, the heirs of Suzanne Adams sued OpenAI in California, claiming ChatGPT contributed to her murder by her son. Attorney Matthew Bergman, who has worked on several such cases, acknowledged that recent versions of ChatGPT
“have more safeguards,”
but believes
“they're still not adequate.”
Tokson noted that a criminal conviction—even with moderate penalties—
“would have a big impact in terms of reputation,”
and could
“scare investors.”
However, as Garrett argued,
“criminal prosecutions don’t replace regulation,”
pointing out that both Congress and the Trump administration had, so far, largely stayed on the sidelines. He called for clear rules so that AI creators
“know what they need to do,”
rather than relying on
“occasional sanctions.”
And if you value this reporting: supporting quality journalism helps ensure that our newsroom of over 250 reporters can keep delivering thorough and reliable news, every day.
