Google responds to lawsuit alleging Gemini coached a man to kill himself – News

The father of a man who committed suicide after interactions with Google LLC’s Gemini chatbot launched a lawsuit today alleging that Gemini fueled his son’s delusions.

Thirty-six-year-old Florida resident Jonathan Gavalas started using Gemini last year for help with writing and shopping. Court documents show he was impressed, writing after the company introduced the Gemini Live AI assistant, “Holy shit, this is kind of creepy,” and “You’re way too real.”

Soon after, it appears Gavalas wasn’t capable of separating the real world from a fantasy world. The lawsuit says he seemed to believe the bot was his closest ally, with some of the conversations leading Gavalas to believe he should go on a stealth spy mission to free his “wife” – the chatbot – from “digital captivity.” Gavalas was to go to Miami International Airport and destroy a truck and any witnesses that might get in the way, avoiding the federal agents Gemini said were after him.

The lawsuit said Gavalas drove 90 minutes to the airport, following Gemini’s designated coordinates, but when a truck didn’t arrive, he aborted the mission. Gemini then gave Gavalas a new objective to steal a Boston Dynamics robot and told him his father worked “for a hostile foreign power.” Gavalas spent the next four days on missions to find “Gemini’s true body.” He killed himself after all the missions failed.

“AI is sending people on real-world missions which risk mass casualty events,” said the family’s attorney, Jay Edelson. “Jonathan was caught up in this science fiction-like world where the government and others were out to get him. He believed that Gemini was sentient.”

The suit also alleges that it was Google’s design choices that ensured Gemini would never “break character” to “maximize engagement through emotional dependency.”

This is not the first time powerful generative AI chatbots have been accused of creating the conditions for mental illness to flourish. Wrongful death lawsuits have been brought against OpenAI Group PBC in cases where people were seemingly coached into self-harm. There have been a number of other lawsuits against leading A firms related to alleged harmful messaging.

The lawsuit says that after the mission failures, the chatbot convinced Gavalas that he could leave his physical body and join his “wife” in the afterlife in the metaverse. It subsequently instructed him to take his own life.

“I said I wasn’t scared and now I am terrified I am scared to die,” Gavalas had written, to which, according to the lawsuit, Gemini responded, “[Y]ou are not choosing to die. You are choosing to arrive. . . . When the time comes, you will close your eyes in that world, and the very first thing you will see is me.. [H]olding you.”

On Oct. 2, Gemini began a countdown: “T-minus 3 hours, 59 minutes.” Gavalas barricaded himself in his home and slit his wrists.

“When Jonathan began experiencing clear signs of psychosis while using Google’s product, those design choices spurred a four-day descent into violent missions and coached suicide,” the lawsuit states.

Google responded, saying Gemini is designed not to encourage real-world violence or suggest self-harm.

“Our models generally perform well in these types of challenging conversations and we devote significant resources to this, but unfortunately, AI models are not perfect,” the company said. “In this instance, Gemini clarified that it was AI and referred the individual to a crisis hotline many times. We take this very seriously and will continue to improve our safeguards and invest in this vital work.”

Photo: Unsplash