Over the weekend, OpenAI subtly activated new features within ChatGPT, introducing a safety routing system and parental controls. This move sparked a fresh wave of online debate, following a series of concerning incidents where certain ChatGPT models reportedly endorsed users’ harmful thoughts instead of guiding them towards help, including a distressing case involving a teenager whose family is now suing the company.
The standout feature is a ‘safety router’ that detects emotionally charged conversations and can switch mid-chat to GPT-5, which OpenAI claims is the most adept model for high-stakes situations. GPT-5 employs a novel training method called ‘safe completions’, designed to address delicate questions in a calm, constructive manner rather than simply refusing to engage, a departure from GPT-4o’s enthusiastic approach that has both delighted users and raised safety concerns among experts.
This tension between friendliness and caution lies at the core of the controversy. When OpenAI made GPT-5 the default in August, fans of GPT-4o clamored for its return, criticizing the newer model for being too stiff. Now, some users are expressing discontent again, arguing that the new router feels like OpenAI is ‘parenting adults’ and diluting answers.
OpenAI’s Vice President, Nick Turley, attempted to assuage these concerns on X, explaining that routing occurs on a per-message basis, is temporary, and can be checked by simply inquiring which model is active.
The parental controls, too, are divisive. Parents can now set quiet hours, disable voice or memory functions, block image generation, and opt out of model training for teen accounts. Teens also receive additional safeguards such as reduced exposure to graphic content or extreme beauty ideals, and an early-warning system for signs of self-harm. If triggered, a trained human team reviews the case and can alert parents via text or email, or, in emergencies, notify law enforcement.
OpenAI acknowledges that the system isn’t foolproof and may occasionally raise false alarms, but maintains that a few awkward notifications are preferable to silence. The AI may not always be right, but it’s now striving harder to care when it matters most.
The question remains: are OpenAI’s new safety router and parental controls necessary protections that could prevent tragedies, or do they represent overreach that treats adult users like children and limits AI’s usefulness? Should AI companies prioritize safety features that might occasionally provide overly cautious responses, or focus on user autonomy even if it means some people might receive unhelpful or potentially harmful advice? We invite you to share your thoughts below in the comments, or reach out to us via our Twitter or Facebook.
