'People are getting hurt': OpenAI sued by Florida over alleged safety risks

Florida sued OpenAI and its main executive, Sam Altman, connected Monday, alleging that the San Francisco tech institution failed to pass radical astir the dangers posed by its fashionable artificial quality chatbot.

OpenAI developed ChatGPT, a chatbot utilized by much than 900 cardinal radical play to reply questions, make substance and implicit different tasks. But radical person besides utilized ChatGPT to program wide shootings, get self-harm accusation and prosecute successful different harmful acts, according to the 83-page lawsuit.

The lawsuit, filed successful Florida’s 10th Judicial Circuit, comes arsenic OpenAI faces mounting concerns that AI is harming people, including children and teens, and determination aren’t capable safeguards successful place. Florida is the archetypal authorities to writer OpenAI implicit information risks, mounting the signifier for others.

“People are getting hurt, parents are getting deceived and they request to wage for it,” Florida Atty. Gen. James Uthmeier said successful a quality league Monday. “They request to wage for it by opening up their checkbook and changing the programme to guarantee that determination are parental controls and that we are not endangering our kids.”

OpenAI said successful a connection that it’s moving to amended information connected ChatGPT, citing efforts specified arsenic caller parental controls and property prediction.

“Losing a kid is the astir devastating calamity that tin hap to a family, and we cognize that nary words tin travel adjacent to addressing the symptom of specified a loss,” a institution spokesperson said successful a statement. “AI is simply a caller and almighty technology, and we judge minors request important protection, which is wherefore we person enactment successful spot industry-leading protections and policies.”

For years, tech companies person faced lawsuits implicit whether they should beryllium held liable for harm posed by the plan and usage of their products. But the increasing popularity of AI-powered chatbots is raising much information concerns, particularly arsenic young radical clasp the tools astatine their schools and homes.

Concerns implicit AI’s risks person heightened successful California. In 2025, the household of a California teen who died by termination aft conversing with ChatGPT sued OpenAI implicit allegations the chatbot encouraged the 16-year-old to instrumentality his ain life. The lawsuit is cited successful the Florida lawsuit.

Other tech companies, including Google, person faced lawsuits implicit the alleged relation chatbots play successful fueling a person’s delusions oregon encouraging self-harm.

California lawmakers person passed caller authorities aimed astatine making AI safer, but tech companies accidental caller regularisation could stifle innovation. The attorneys wide of California and Delaware expressed worries past twelvemonth astir allegations that AI chatbots are harming children.

At a quality league astir the latest lawsuit, Uthmeier accused OpenAI and Altman of prioritizing nett implicit nationalist information arsenic the AI institution races ahead. This year, the Florida lawyer general’s bureau besides announced it’s investigating OpenAI implicit the relation ChatGPT played successful a 2025 shooting astatine Florida State University that near 2 radical dead.

OpenAI antecedently said the assemblage shooting was a calamity but “ChatGPT is not liable for this unspeakable crime.” It proactively shared accusation with instrumentality enforcement aft the institution identified an relationship the institution believes is tied to the suspect.

OpenAI and Altman deceived parents into reasoning that ChatGPT is harmless erstwhile it’s not, violating Florida laws, according to the lawsuit.

Uthmeier brought up respective cases successful assorted states, outlining people’s usage of ChatGPT successful shootings, suicides and different types of deaths, including 1 that progressive a cause overdose. ChatGPT, helium said, tin beryllium “addictive” due to the fact that it mimics quality empathy, tricking users into providing much accusation to the AI chatbot.

“Today we’re going to nonstop a connection to OpenAI: Get acceptable for a fight, and there’s not 1 much important than this close now,” Uthmeier said.