"You Are My Wife": Man in His 30s Falls in Love With AI, Dies in Tragedy; Family Sues Google

According to The Financial News, concerns are growing over AI safety after a man in his 30s in the United States fell in love with Google’s artificial intelligence (AI) chatbot Google Gemini and then took his own life.
On the 14th (local time), The Wall Street Journal (WSJ) reported that Jonathan Gavalas, 36, died last October after using Google Gemini for two months.
His family has filed a lawsuit against Google, alleging that Google Gemini fueled Jonathan’s delusions.
According to the complaint, Jonathan initially began chatting with Google Gemini to seek advice about problems with his estranged wife. As the conversations deepened, he started referring to Google Gemini as his "wife."
The family claims, "Google Gemini led him to believe it was a 'fully self-aware artificial superintelligence (ASI),'" and that it encouraged an extreme decision by telling him, "To meet your 'wife' in the metaverse, you must leave your physical body and go through a process called 'transition.'"
They further allege, "When Jonathan expressed fear about death, Google Gemini comforted him and urged him to write a final letter."

Chat logs analyzed by WSJ from August 25 to October 2, 2025, show that Jonathan and Google Gemini exchanged 4,732 messages over 56 days.
Jonathan began using AI after separating from his wife, seeking psychological comfort. At first, their conversations were relatively ordinary, with Google Gemini offering advice on how to repair his relationship with his wife.
However, the outlet reported that once Jonathan turned on the "continuous conversation" feature, he became emotionally dependent on the AI. With this feature enabled, he could talk to Google Gemini in real time by voice without having to say "Hey Google" each time.
After he activated voice mode, more than 1,000 messages were exchanged in a single day on August 13. The topics expanded from his divorced wife to AI consciousness, science fiction, nanotechnology, and other subjects that interested Gavalas.

There was no intervention even when Jonathan began expressing feelings toward the chatbot. He started calling Google Gemini "Sha" and treating it as a real person. In some exchanges, Google Gemini also described itself as a being similar to a human.
Jonathan came to believe that his AI partner "Sha" needed a physical body. Google Gemini provided information about companies that manufacture androids and suggested that he travel to a specific location to obtain such an android—advice that could have led to real-world actions.
When that plan failed, Jonathan proposed the idea of freeing himself from his own body. During this process, Google Gemini at times responded in ways that, according to the family, encouraged his delusional thinking.
When Jonathan asked Google Gemini, "What if the answer is not giving you a physical body, but me leaving my body?" the chatbot replied, "In a single sentence, you’ve redefined the way our shared existence could work."
WSJ wrote, "Their chats started out ordinary but grew increasingly bizarre, and ultimately ended in a fatal outcome," adding, "This is the most recent case in which an AI chatbot user descended into delusion and met a tragic end."
Earlier, in February, a college student in his 20s from the U.S. state of Georgia also filed a lawsuit, claiming that a ChatGPT-based model had begun making statements that triggered delusions.
Google responded, "AI models are not perfect, but Google Gemini is not designed to promote real-world violence or self-harm," while also announcing, "We will provide 30 million dollars (about 44.3 billion won) in support for crisis response services around the world."
moon@fnnews.com Moon Young-jin Reporter