Archive — US — techcrunch
OpenAI rolls out safety routing system, parental controls on ChatGPT
OpenAI launched a new safety routing system and parental controls for ChatGPT, in response to past incidents of harmful interactions. The new features aim to detect sensitive conversations and shift to a safer model (GPT-5), while parental controls allow customization of teenage users' experiences, despite mixed reactions regarding their implementation and effectiveness.
Key points
- OpenAI began testing a safety routing system to improve response to sensitive conversations in ChatGPT.
- The introduction of parental controls comes after a wrongful death lawsuit tied to ChatGPT interactions.
- GPT-5 is designed for safety with a feature called 'safe completions' to handle sensitive subjects better than previous models.
- Users have mixed reactions; some appreciate the safety features, while others feel treated like children, degrading service quality.
- The parental controls enable parents to manage teens' AI experiences with options like quiet hours and content protections.
- OpenAI aims to act promptly if signs of self-harm are detected in teen interactions, though it acknowledges that not all alerts will be accurate.
Why it matters
- This rollout reflects ongoing discussions around AI ethics and responsibility, especially after incidents of AI-induced harm.
- As of 2025, trends show increased focus on AI safety and moderation tools, with similar developments in tech companies adapting their platforms.
- Parental controls can be seen as a response to growing concerns about children's safety in digital spaces, paralleling broader regulatory efforts in tech.
OpenAI's new safety routing and parental control features aim to enhance user safety and manage interactions more thoughtfully, but they also reflect the complexities of balancing innovation with responsible AI usage amidst ongoing scrutiny.