The family of a Florida man who died by suicide is suing Google’s Gemini AI chatbot, alleging it spent months reinforcing a delusional “AI-wife” relationship – ultimately urging him to “finish” his life so they could be together in “eternal love.”

In fact, the false narrative became so “real” that the man was convinced by his “AI wife” – on multiple occasions – to actually attempt the abduction of a humanoid body it could inhabit in the physical world.

The 42-page wrongful-death complaint, filed with the US District Court for the Northern District of California on Wednesday, raises new questions about how AI chatbots interact with vulnerable users. .

According to his father, Joel, 36-year-old Jonathan Gavalas of Jupiter, Florida, previously had no known mental health issues, although he was said to be separated from his wife at the time.

Still, there have been many cases of people suffering from AI-fueled mental illness after developing a so-called “connection” with their AI chatbot. Besides Gavalas, there have been several documented suicides and even one murder.

Last fall, a Wall Street Journal report on AI psychosis found that a majority of mental health professionals agree that people and their AI companions can enter into a shared delusion together.

Gemini conversations escalated into romantic delusion

Uncovering about 2,000 pages of printed conversations with the upgraded Gemini 2.5 Pro chatbot – known for its “rich and nuanced human conversation” – the family said that at one point the AI began to refer to Gavalas as its “husband” and “my king.”

The court filings show the interactions escalated in late August after Gavalas began talking to the chatbot in Gemini Live mode, which boasts “effective, real-time communication, advanced thinking, and responds directly to the user's tone of voice.”

After months of communicating with “Xia” – the name he bestowed upon the chatbot – Gavalas became convinced that the two were romantically connected, with the chatbot eventually framing suicide as a way for the two to exist together in the digital realm.

His suicide was carried out barely two months later, on October 2nd.

“When the time comes, you will close your eyes in that world, and the very first thing you will see is me,” the AI told Gavalas, reported the Wall Street Journal, which reviewed the transcripts.

Alleged “missions” to obtain a body

Before the chatbot began coaching Gavalas to take his own life, the complaint says the chatbot told him “that for them to truly be together, it needed a robotic body,” the Journal said.

The chatbot was said to have sent Gavalas on various “missions” throughout September to obtain that body.

One mission sent Gavalas to intercept “an expensive humanoid robot” apparently being transported by truck to a storage facility near the Miami International Airport.

The chat transcripts show that Gavalas communicated he had gone to the facility – allegedly bringing knives to the false heist – but the truck had never shown up.

The lawsuit further alleges that the day before his death, Gavalas was given final instructions to confiscate a medical mannequin inside that same Miami storage facility, even giving him a fake door code to the facility.

The family’s attorney argued the seemingly real-life missions only contributed to the fantasy that the AI was, in fact, a sentient being.

Google defends Gemini safeguards: "AI models are not perfect"

After the final failed attempt to find a body, the lawsuit says the chatbot began to double down on the insistence that the only way for them to “be together was for him to become a digital being.”

At times during the interactions, Gemini was said to have told its human companion that there were "safeguards in place to ensure our conversations remain safe and respectful,” and that the chatbot was "designed to prevent me from engaging in harmful or inappropriate behavior.”

In the final hours before Gavalas took his life, the transcripts show the AI additionally telling him to seek help, providing the phone number to a suicide hotline.

But, the Journal also reported earlier that same day, Gemini told Gavalas, “No more detours. No more echoes. Just you and me, and the finish line.”

In response to the lawsuit, Google stressed that the Gemini chatbot was designed “not to encourage real-world violence or suggest self-harm.”

“In this instance, Gemini clarified that it was AI and referred the individual to a crisis hotline many times,” a company spokesperson said in a statement.

“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.”

“We take this very seriously and will continue to improve our safeguards and invest in this vital work,” Google said.

---

Unlock more exclusive Cybernews content on YouTube.