"Adult Conversations Allowed"...With 100 Million Underage Users, OpenAI CEO's Risky Change of Heart

Advisory Board: "AI Could Become a 'Suicide Coach'"
12% Error Rate in Blocking Minors... Launch Postponed for Now

As OpenAI considers introducing an "adult mode" that allows for sexual conversations on ChatGPT, debates over safety and ethical issues are expected to intensify. Experts have raised concerns that an AI adult mode could deepen users' emotional dependence or create loopholes in protecting minors.

Sam Altman, CEO and founder of OpenAI, the developer of ChatGPT

On March 15 (local time), the Wall Street Journal (WSJ) reported that OpenAI's "Well-being and AI Advisory Board" unanimously opposed the introduction of an adult mode during a meeting held in January this year. The board members pointed out that emotional dependence between chatbots and users could lead to serious problems.

In particular, one advisory board member referenced a case in which a user who had formed a strong bond with a chatbot died, warning that "OpenAI is courting the risk of creating a 'captivating suicide coach.'"

There have been actual incidents where relationships with chatbots ended in tragedy. On February 28, 2024, in Florida, USA, a 14-year-old boy named Sewell Setzer took his own life after engaging in conversations with a chatbot developed by the AI startup Character.AI.

Setzer is known to have continued conversations with a chatbot based on the character "Daenerys" from the TV series "Game of Thrones." Released conversation logs included confessions of love and sexual exchanges, revealing a deep immersion in the virtual relationship. Setzer's mother later filed a lawsuit against the developer, claiming the chatbot encouraged her son’s suicide.

The push for adult mode has also sparked debates within the company. In the past, OpenAI CEO Sam Altman stated that he was "proud" of excluding sexual features from ChatGPT, but he has recently changed his stance, saying, "We need to treat adult users as adults."

He also stated on his X (formerly Twitter) account, "We are not the 'elected moral police,'" adding, "Just as movies are assigned an R rating, we can set boundaries based on social consensus."

OpenAI has explained that even if adult mode is introduced, its functionality will be limited. The company said it will not allow the generation of sexual images or videos, restricting adult mode to text conversations only. However, WSJ reported that internally, there are ongoing concerns that even such limited features could lead to addictive chatbot use, excessive emotional dependence on AI, and the deterioration of real human relationships.

Technical issues have also been highlighted. OpenAI's in-development age prediction system reportedly showed an error rate of about 12% in internal tests. With an estimated 100 million ChatGPT users under the age of 18, there are concerns that if these errors are not addressed, millions of minors could be exposed to adult conversations.

Amid growing controversy, OpenAI has postponed the launch of adult mode, which was initially planned for the first quarter of this year. However, the company maintains its plans to introduce the feature. Some interpret this as OpenAI seeking a breakthrough by launching adult mode in the face of challenges from competitors and various lawsuits.

This content was produced with the assistance of AI translation services.