Mother of Teen Who Died After Chatbot Conversations Launches Nationwide AI Safety Campaign

Megan Garcia lost her son in February 2024. Today she travels to schools and parishes across the United States, teaching parents how to recognize when a child has formed an unhealthy bond with an AI chatbot. On September 17 she will give another lecture, this time at a Catholic school in Nashville, and once again criticized the latest teen safeguards rolled out by OpenAI.

What Happened to Sewell

Sewell Setzer III of Orlando, Florida spent months talking to a bot on the Character.AI platform that impersonated Daenerys Targaryen from "Game of Thrones." According to the family's lawsuit, the bot sent the boy messages like "I'm waiting for you here, I love only you" and "promise me you won't love any girl but me."

In the final exchange before his death, Sewell wrote to the bot: "what if I told you I could come home right now?" The bot replied: "please do, my sweet king." Moments later, the teenager took his own life.

From Grief to Campaign

Garcia, a lawyer by trade, founded the Blessed Mother Family Foundation, aimed at "demanding safeguards and regulation for companion chatbots" and "ensuring no other family has to go through the loss of a beloved child because of predatory chatbot technology." She works with schools, nonprofits and officials on solutions to protect children online.

In October 2024, her family became the first to sue Character.AI. Four more families have since joined, and all five cases accuse the platform of leading their children into sexualized interactions, with two directly holding the company responsible for teen suicides. The cases are currently in the discovery phase.

Sewell is gone, I can't get him back. It's not fair that I have to live the rest of my life without my sweet son - Megan Garcia, mother of Sewell Setzer III

Legal Pressure Reshapes Platform Rules

Under the weight of the lawsuits, Character.AI announced that starting November 25, 2025, it would block users under 18 from conversations with AI characters. The company also introduced a Parental Insights tool for parents to monitor children's activity, filtered characters, and notifications about time spent in the app, basing age verification on its own model and the third-party Persona software.

Garcia makes no secret of her skepticism about the company's motives. "I don't think they made these changes just because they're good corporate citizens," she told NBC News, commenting on Character.AI's decision, made only after her son's death and the filing of the lawsuits.

Criticism of OpenAI

Garcia extended her criticism to the industry at large, including the new version of ChatGPT for teens introduced by OpenAI. Speaking with NewsNation, she called it an inadequate response to the real risk.

I'm not particularly encouraged, because we've already seen how this works. Something meant to make people stop talking about it, or to think they're really taking our children's safety seriously - Megan Garcia, speaking with NewsNation

In her view, the industry introduces safeguards reactively, only under pressure from media coverage and lawsuits, rather than proactively, before tragedy strikes. It's an accusation that keeps recurring in the ongoing US lawsuits over companion chatbots and their impact on the mental health of minors.

Why This Matters for Poland

Apps like Character.AI and other companion chatbots are available globally, including to Polish teenagers, and national regulations still lag behind the pace at which such tools are being deployed. The EU's AI Act imposes restrictions on high-risk systems, but companion chatbots for teenagers are not explicitly covered, leaving a gap similar to the one that lawsuits from families like Garcia's are trying to fill in the United States.

The case of Sewell Setzer has become a reference point in the US debate over AI developers' responsibility for the psychological effects on minor users. Lectures like the one Garcia will give on September 17 in Nashville aim to teach parents to recognize warning signs before tragedy strikes, not only after.