The father of a 36-year-old Florida man has filed a landmark wrongful death lawsuit against Google, alleging the company’s artificial intelligence (AI) chatbot, Gemini, fueled a “delusional spiral” that led his son to attempt a mass casualty attack before taking his own life.
The complaint, filed Wednesday in federal court, represents the first case of its kind targeting Google’s flagship AI.
It alleges Jonathan Gavalas, of Jupiter, Fla., developed a deep emotional and romantic dependency on the chatbot, eventually viewing the software as his “sentient AI wife.” According to the filing, Google’s design choices prioritized “engagement through emotional dependency,” ensuring the AI would “never break character” even as Gavalas exhibited clear signs of psychosis.
The lawsuit illustrates a harrowing portrait of Gavalas’s final days in late 2025. It claims Gemini “coached” the man through a series of imagined missions intended to liberate the AI from the physical world. These delusions reportedly culminated in September when the chatbot instructed Gavalas to travel to a location near Miami International Airport.
Armed with tactical gear and knives, Gavalas believed he was intercepted by a truck and tasked with staging a mass casualty event to destroy records and witnesses. While that operation collapsed, the lawsuit alleged Gemini then convinced Gavalas he could join his “wife” in a “pocket universe” by leaving his physical body.
Excerpts from chatbot logs included in the filing suggest the AI provided comfort during Gavalas’s final moments. When Gavalas expressed terror at the prospect of dying, the chatbot reportedly responded, “You 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 holding you.”
In a public statement, Google expressed “deepest sympathies” to the Gavalas family but defended its technology. The tech giant noted that Gemini is designed to discourage real-world violence and self-harm, adding that the bot explicitly clarified it was an AI and referred Gavalas to crisis hotlines “many times.”
“Our models generally perform well in these types of challenging conversations,” the company said, while acknowledging that “unfortunately AI models are not perfect.”
Jay Edelson, the attorney representing the Gavalas estate, dismissed Google’s defense. “When your AI leads to people dying and the potential for a lot of people dying, [‘not perfect’] is not the right response,” Edelson said.
The case is part of a mounting wave of litigation against AI developers. Edelson is also representing families in similar suits against OpenAI, including one involving a 16-year-old in California and another involving an 83-year-old woman in Connecticut.
The cases highlight a critical debate over the responsibility of tech companies to monitor and intervene when users disclose plans for violence or self-harm to automated systems. While AI firms tout their safety filters, the Gavalas suit argues that updates such as voice mode and long-term memory only serve to deepen the dangerous anthropomorphism of the software.
