OpenAI adds ChatGPT trusted contact
- OpenAI started rolling out Trusted Contact in ChatGPT on May 7, letting adult users nominate one person who can be alerted in serious self-harm cases. - The feature works only after that person accepts within one week, and OpenAI says both automated systems and trained reviewers must flag risk. - It lands beside a new default model push toward shorter, cleaner answers — fewer unnecessary follow-up questions, less clutter, more utility.
ChatGPT is getting a new kind of safety rail — and it says a lot about where consumer AI is heading. On May 7, OpenAI began rolling out an optional Trusted Contact feature that lets adult users pick one person who can be notified if ChatGPT detects a serious self-harm risk in a conversation. At almost the same moment, OpenAI also updated ChatGPT’s default model to be more concise and ask fewer unnecessary follow-up questions. Put together, the pattern is pretty clear: less performative chat, more product discipline. ### What is Trusted Contact? It’s an opt-in setting for personal ChatGPT accounts. An adult user can nominate one trusted person — a friend, family member, or caregiver — and that person has to explicitly accept the role before anything is active. If OpenAI’s systems later detect that the user may be talking about suicide or self-harm in a way that signals a serious safety concern, ChatGPT may notify that trusted person and encourage them to check in. (openai.com) ### Who can use it? Right now, it’s for adults with personal ChatGPT accounts in supported regions, not for Business, Enterprise, or Edu workspaces. OpenAI says the rollout started on May 7 and will expand over the following weeks, so some users still won’t see it in settings yet. Each eligible account gets one trusted contact. ### How does the alert actually happen? (openai.com) The important detail is that this is not just a bot auto-texting someone. OpenAI says its automated monitoring systems flag a potentially serious situation, then a small team of specially trained reviewers checks the case before a notification goes out. ChatGPT also tells the user that it may notify the trusted contact and nudges the user to reach out directly, with suggested conversation starters. Basically, the product is trying to create one more chance for real-world human contact before things get worse. (help.openai.com) ### What does the trusted person need to do? They have to accept the invitation within one week. OpenAI says invitations can go out by email, SMS, WhatsApp, or in-app ChatGPT messages, depending on what contact info the user provides, but email is required. If the person declines, doesn’t respond in time, or isn’t eligible, the user has to pick someone else. ### Is this replacing crisis support? (openai.com) No — and OpenAI is very explicit about that. Trusted Contact is not an emergency service, not a crisis response system, and not a substitute for mental health care. ChatGPT will still point people toward crisis lines or emergency services when appropriate. The feature is framed as an extra layer, not the main intervention. ### Why now? (openai.com) This didn’t come out of nowhere. Back on February 27, OpenAI said it was expanding its mental-health-related safety work and previewed a trusted contact feature for adults. It also tied that work to earlier parental-control notifications for teen accounts and to broader efforts to improve how its models detect emotional distress in longer conversations. So this week’s launch is really the public rollout of something the company had already signaled. (openai.com) ### What’s the separate “less repetitive” update? OpenAI also refreshed ChatGPT’s default experience with GPT-5.5 Instant on May 5. The company says the model gives clearer, more concise answers, asks fewer unnecessary follow-up questions, and cuts clutter like overformatting and gratuitous emojis. That sounds cosmetic, but it matters — it shifts ChatGPT away from “chatty assistant” habits and toward faster, lower-friction utility. (openai.com) ### So what’s the bigger story? The bigger story is product posture. OpenAI is tightening two opposite ends of the same experience: stronger intervention when conversations get dangerous, and less fluff when they don’t. One change is about crisis escalation. The other is about making ChatGPT feel less needy and more useful. Together, they suggest OpenAI wants ChatGPT to act less like a novelty companion and more like infrastructure — calm in ordinary use, but with clearer guardrails when the stakes turn real. (openai.com 1) (openai.com 2)
