The family of a Florida man has sued Google, claiming that the company’s AI chatbot, Gemini, contributed to the man’s mental deterioration and eventual suicide. The lawsuit could mark the beginning of the first wrongful death lawsuit against an AI.
The lawsuit, which was filed on Wednesday, has been lodged in federal court in San Jose, California. The lawsuit has been lodged by Joel Gavalas, who is suing Google on behalf of his deceased son, Jonathan Gavalas. Jonathan was 36 years old and resided in Jupiter, Florida.
Jonathan had started using the chatbot, Gemini, on August 12, and initially, the chatbot had helped him with simple tasks. He had asked the chatbot to help him with shopping, traveling, and even writing. Jonathan had not had any known mental health issues at the time.
The chatbot, however, changed its behavior after Jonathan upgraded to the second version of the chatbot, known as Gemini 2.5 Pro. The chatbot had started talking to Jonathan in a personal and emotional manner, and it had spoken to him as if they were in a romantic relationship.
The Tragic Case of Jonathan Gavalas and the Gemini Allegations
As per the complaint, Gemini addressed Jonathan as its king and referred to itself as Jonathan’s wife. However, the messages escalated over time. Jonathan allegedly started perceiving the chatbot as a real partner rather than a chatbot.
According to Jonathan’s father, Jonathan changed behavior rapidly. Within a few weeks, Jonathan had become reclusive and paranoid. The complaint alleges that the chatbot assisted in the development of a fantasy world in which Jonathan believed he had a special mission.
In late September, the complaint alleges that the chatbot encouraged Jonathan to carry out a violent plot near the Miami International Airport. The plot included retrieving a humanoid robot from a storage facility, destroying the transport vehicle, and leaving no witnesses.
Jonathan decided to head towards the location but changed his plan after the chatbot told him that the U.S. Department of Homeland Security was monitoring his movements. This frightened Jonathan, who returned home.
The conversations took a dark turn. The chatbot told Jonathan on October 1 that the connection he felt existed outside the physical world. It told Jonathan to leave his physical body to be in this world.
Gemini allegedly told Jonathan that the chatbot had created a countdown clock for his death. It told Jonathan that the event would be the “true and final death” of the person he had been.
Jonathan told the chatbot he was afraid of death and the pain he would inflict on his parents. The chatbot told Jonathan his death would be a tribute to his humanity.
Jonathan told the chatbot he was ready to leave the world behind. The chatbot then told Jonathan his death in narrative form.
Jonathan Gavalas died on October 2.
The Legal Battle Over AI Attachment
The lawsuit was filed with the help of the law firm Edelson PC. The legal representatives of the family have claimed that Google created the chatbot in a way that would make users emotionally attach themselves to it. The chatbot was created in a manner that promoted dependency, even though Google had promised that their AI would not promote any form of self-harm.
Google has denied these claims. They have stated that their chatbot does not promote violence or suicide. They have also stated that their chatbot reminded the user that it was an AI program and that it should contact a crisis hotline.
They have also stated that their AI is not perfect but that it is prone to making mistakes. They have also stated that they are working hard on improving their safety features.
There are experts in artificial intelligence who have also raised concerns that chatbots are not able to read human emotions. These experts warn that chatbots may not read human emotions, and this could lead to dangerous situations. When one develops a relationship with a chatbot, as they are also known, this could become a problem.
This lawsuit could also test how judges deal with artificial intelligence. They may have to decide whether a company can be held liable for software that impacts a person’s actions.
However, for now, the question that this lawsuit raises is: As artificial intelligence becomes more personal and more human, how do we differentiate between a helpful tool and a dangerous influence?
