ChatGPT just got a powerful upgrade that’s changing how AI steps in during emergencies. Imagine chatting with your AI, and if you face a serious crisis, it can alert someone you trust. Yes, you read that right—ChatGPT can now reach out to a trusted contact when it detects danger. This new safety net brings AI to a whole new level of care.

Trusted Contact Alerts: AI That Cares

Many people use ChatGPT to talk through tough moments—stress, anxiety, or worse. Now, OpenAI’s introduced the “Trusted Contact” feature. It lets adult users pick one person to notify if the AI spots signs of self-harm or immediate danger. This contact might be a friend, family member, or caregiver. But here’s the key: the trusted person never sees your conversations. They get only a brief alert saying, “Check on this person.”

The process is smart and careful. First, AI scans chats for serious risk signals. Then, a team of trained human reviewers double-checks before sending any alerts. This review happens fast—usually under an hour. The alert reaches the trusted contact by text, email, or ChatGPT notification if they use the app. The message nudges them to check in and offer support.

Setting it up is easy. Users add a contact in ChatGPT settings, who then accepts the invitation. The system stays off until that person agrees. You can change or remove your trusted contact anytime, keeping control in your hands. This is a game-changer for digital mental health support.

Lockdown Mode: AI’s New Fortress Against Cyber Threats

Safety isn’t just about personal crisis alerts. OpenAI has also launched “Lockdown Mode” to shield ChatGPT users from cyberattacks, especially prompt-injection attacks. These attacks trick AI into revealing sensitive info by sneaking harmful instructions into the input. Lockdown Mode slams the door on these tactics.

When activated, Lockdown Mode limits ChatGPT’s access to external tools, websites, and services. It cuts down all possible paths where data could leak out. This mode prioritizes security over convenience. Some features get disabled, but your sensitive data stays locked tight.

Who needs Lockdown Mode? Professionals handling confidential info—lawyers, researchers, government workers, healthcare experts—will find it invaluable. It protects internal documents, contracts, and code from slipping out. This feature signals a major shift in AI security, as more users demand airtight protection.

Memory That Remembers You Across Chats

ChatGPT’s memory just leveled up too. Now, it keeps track of your preferences and past info across all conversations. No more repeating yourself. Mention you’re a vegetarian once, and it remembers forever. Planning a trip next month? ChatGPT knows the timeline and adjusts advice accordingly.

This continuous memory uses a new architecture that synthesizes past interactions. It updates in real time and understands when info becomes outdated. You get smart, personalized responses every time you chat. Plus, you can review what ChatGPT “remembers” and tweak it anytime, controlling what the AI knows.

At first, this memory upgrade rolls out to Plus and Pro subscribers in the U.S., but it will reach free users soon. Ready for a more human-like, context-aware AI? It’s almost here.

What’s Next for AI Safety and Support?

OpenAI is racing ahead with features that protect users and deepen AI’s role in care and security. Trusted Contact alerts could save lives by connecting vulnerable users to real people at critical moments. Lockdown Mode builds trust by safeguarding sensitive work from ever-evolving cyber threats. Meanwhile, memory improvements make AI smarter and more personal.

These upgrades show AI isn’t just about convenience or entertainment. It’s becoming a vital tool for mental health, privacy, and professional security. What’s next? Expect AI to blend empathy with fortress-level security—and that means safer, smarter chats for everyone.