Despite lawsuits and ongoing security concerns: Character.AI co-founder switches to OpenAI

Character.AI co-founder Noam Shazeer moves to OpenAI. Most recently, he worked as Vice President of Engineering at Google, where, among other things, he was responsible for the further development of the Gemini chatbot. In a post on However, the personnel does not only cause enthusiasm. Above all, allegations regarding safety and child protection cause criticism.

Mother makes serious accusations

As Futurism reports, this is Shazeer’s second farewell to Google. Together with his colleague Daniel de Freitas, he left the group for the first time in 2021 to found Character.AI. As the Wall Street Journal reported, her ex-employer is said to have previously refused to publish a chatbot developed by the two of them – due to security concerns. By March 2023, the startup reached a billion dollar valuation, although it did not record any revenue at that time. In August 2024, the two returned to Google: The tech company is said to have invested $2.7 billion in Character.AI to bring Shazeer and de Freitas back.

Since then, the AI ​​startup has been criticized several times. In October 2024, a mother named Megan Garcia filed a lawsuit against Character.AI and Google. She claimed her son took his own life after developing parasocial relationships with several of the platform’s chatbots over a period of months. His last conversation is said to have taken place with a chatbot run by the character Daenerys Targaryen Game of Thrones was inspired and urged the teenager to “come home to (him) as soon as possible”. At the time of the suicide, Shazeer and de Freitas were still at the top of the company.

Questionable chats with Jeffrey Epstein

Before the trial began, Garcia and the two companies reached an out-of-court settlement. The agreement came a few months after Character.AI decided to restrict access to the platform for users under the age of 18. As part of this measure, the company stated that it had developed an internal tool to identify underage users based on their chat histories. The company also entered into a cooperation with a third-party provider for age verification using official identification documents. However, it remains questionable whether these measures are sufficient to protect children and young people.

The type of chats also repeatedly cause criticism: As Futurism reported in March 2026, chats had again appeared on the platform in which users could “meet” the deceased sex offender Jeffrey Epstein. The Bureau of Investigative Journalism had already reported on such bots in October 2025. According to Futurism, access was only possible via accounts assigned to adult users. Nevertheless, the sometimes disturbing introductory texts are said to have been readable even with a young person’s account.

Switching to OpenAI causes criticism

According to the two founders, Character.AI was deliberately designed to be open. “Our goal has always been to bring something to market and let users decide what they think it is good for,” Shazeer said in a December 2023 podcast interview, according to Futurism. Now he is moving to OpenAI – a company that is also facing lawsuits related to suicides and so-called AI psychoses. Early last year, Futurism interviewed Garcia about the Character.AI founders’ return to Google, which came just months after their child’s suicide. “In my opinion, these two gentlemen should not have the right to continue developing products for people, let alone children – especially children,” said the mother.