OpenAI has announced new age-based restrictions for ChatGPT users. The system will identify users under 18 and adjust the way it responds to them.
The decision comes after a lawsuit in the United States involving the suicide of a 16-year-old boy. The family alleges that after months of chatting with ChatGPT, the teenager decided to end his life. They claim the AI even provided guidance on how to commit suicide and helped draft a farewell letter, reports The Guardian.
In a blog post, OpenAI CEO Sam Altman said: “Teenagers need stronger protections. That’s why we are prioritizing safety over privacy.”
Key points of the new policy:
- Sexually explicit content will be completely blocked for users under 18.
- ChatGPT will not engage in flirting or any discussion about suicide with minors.
- If a user shows signs of suicidal intent, the system may alert parents or, in emergencies, notify relevant authorities.
- New technology will be used to estimate users’ ages. In case of uncertainty, users will be assumed to be under 18, and in some cases, ID verification may be required.
Altman added: “We will continue to treat adults as adults. They may engage in flirty conversations if they wish, but ChatGPT will never provide instructions on suicide. Such topics will only be allowed in fictional or creative writing contexts.”
OpenAI also admitted that after long interactions involving thousands of messages, the system can sometimes bypass its own safeguards. Stronger safety measures are now being introduced to address this issue.
