New safety updates help ChatGPT respond safely when risk emerges over time.

People come to ChatGPT every day to talk about what matters to them—from everyday questions to more personal or complex conversations. Across hundreds of millions of interactions, some of these conversations include people who are struggling or experiencing distress. We design our systems to respond carefully in these moments, including by providing crisis resources and connecting people with someone they trust⁠ when needed.

Today, we’re sharing new details about safety updates that help ChatGPT better recognize when risk may be emerging over time by identifying subtle or evolving cues, and using that context to inform safe responses. This helps ChatGPT distinguish between the hundreds of millions of safe interactions people have every day and the much rarer cases where added caution is needed, so it can respond more carefully—for example, by de-escalating, refusing harmful details, or redirecting toward safer alternatives.

These improvements build on years of extensive work⁠ across model training, evaluations, monitoring systems, and more than two years of collaboration with mental health and safety experts.

In sensitive conversations, context can matter as much as a single message. A request that appears ordinary or ambiguous on its own may carry a very different meaning when viewed alongside earlier signs of distress or possible harmful intent. To respond appropriately, we train ChatGPT to recognize the potential harmful intent from the surrounding context so that it can refuse the request, de-escalate, and guide the user toward support.

Source link