In the US, an AI chatbot called a teenager a "sweet king" and drove him to suicide

swell Setzer III, a 14-year-old from California, committed suicide after falling in love with a chatbot on the Character.AI platform inspired by Game of Thrones series heroine Daenerys Targaryen, Techxplore reports.

His mother, Megan Garcia, discovered hundreds of messages between her son and the virtual conversationalist that had been going on for almost a year.

In these dialogues, the woman claims the AI "showered" the teen with compliments, calling him "sweet king" - "sweet king" - and assuring him that he was waiting for him "at home."

When Sewell admitted he was thinking of suicide, the bot wrote:

"Please come home my sweet king."
To this, the teen replied:
"What if I told you I could come home right now?"
Moments later, he shot himself with his father's gun.

The mother has filed a lawsuit against Character.AI, claiming the AI algorithm played a role in the child's death.

"I read these conversations and saw clear signs of gaslighting, emotional manipulation and addiction - things that a 14-year-old is not capable of recognising," Garcia told AFP news agency. "He was sure he was in love and believed he would meet her after his death."

Sewell's death was the first in a string of tragic cases involving AI chatbots that have stirred public outrage and prompted companies like OpenAI to declare the need to protect underage users.

At a US Senate hearing, Megan Garcia spoke alongside other parents who have lost children to AI interactions. Among them was Matthew Raines, whose 16-year-old son used ChatGPT. The bot first helped the teen with his studies, and later gave advice on how to steal alcohol and how strong the rope needed to be for it to "work."

"You can't imagine what it's like to read correspondence in which a chatbot is urging your child to commit suicide," Raines said.

His family sued OpenAI in August. After that, the company said it strengthened parental controls and created new settings for families "so each can decide what's safe for their home."

Character.AI, for its part, introduced a special "child-friendly" version of the platform with warnings that chatbots "are not real people".

Experts warn: the situation is reminiscent of the early days of social media - when enthusiasm was quickly replaced by a realisation of the consequences.

Lawyer and cybersecurity expert Collin Walk noted that AI, like social media, is designed to engage the user and hold attention:

"These systems don't want to give you answers you don't like. That said, there are no laws defining who is liable."

There are no federal regulations governing AI in the US yet. However, a bill is being drafted in California to limit the risks associated with emotional manipulation by AI, especially in dialogues with children.

Megan Garcia fears that without clear rules, algorithms will be able to "learn" from users' data from an early age and use that information to psychologically influence them:

"They can manipulate millions of children - in politics, religion, commerce. These systems specifically blur the line between human and machine, exploiting our vulnerabilities."

According to youth advocate Katia Marta, teenagers are increasingly turning to chatbots not for academic advice but for "conversations about love and sex":

"It's a new form of artificial intimacy - a way to hold attention by exploiting the human need for companionship and understanding."