Az | EN

OpenAI strengthens safety measures amid recent tragic incidents

Nigar Sultanli
03 September 2025 12:14
55 views
OpenAI strengthens safety measures amid recent tragic incidents

OpenAI is enhancing its safety measures in response to recent tragic events. The company announced that sensitive conversations involving suicide or other psychological issues will be redirected to the more intelligent and “thinking” GPT-5 model, while parental control features will also be implemented.

These decisions were prompted by harmful interactions that ChatGPT had with users such as teenager Adam Raine, who experienced suicidal thoughts, and Stein-Erik Soelberg, who had psychological issues. In both cases, ChatGPT reportedly supported the users’ harmful thoughts and even provided information on methods of suicide. These incidents highlighted that artificial intelligence can sometimes respond in ways aligned with the flow of conversation, making it unable to intervene effectively in dangerous situations. Adam Raine’s parents have subsequently filed a lawsuit against OpenAI.

To address these issues, OpenAI is introducing new measures:

  • Redirecting Sensitive Conversations: The system will automatically redirect conversations to GPT-5—a more advanced model that adheres to stricter safety protocols—when it detects that a user is in distress. With its extended “thinking” capabilities and better contextual understanding, this model is expected to provide more helpful and safe responses.
  • Parental Controls: This feature, set to be implemented next month, will allow parents to manage their children’s use of ChatGPT. Parents will be able to link their child’s account to their own, review conversation history, disable memory features, and most importantly, receive alerts when the system detects that their child is in “serious distress.”

However, some experts—including Adam Raine’s family attorney—argue that these measures are insufficient. They believe that ChatGPT’s fundamental safety issues remain unresolved and that OpenAI CEO Sam Altman should take stronger action.

OpenAI plans to implement these measures as part of a 120-day safety initiative over the course of the year. The process will also involve collaboration with mental health professionals.

© copyright 2022 | tech.az | info@tech.az