November 2, 2026

OpenAI Addresses AI Safety Concerns and Enhances ChatGPT Privacy Features

AI Safety Concerns and Government Hack

OpenAI’s Chief Global Affairs Officer, Chris Lehane, discussed the investigation into a security incident involving an OpenAI hack on an Australian government Medicare portal. In an interview with Lawrence Jones, Lehane detailed the context of the July breach and shared findings from their ongoing review. Additionally, he responded to a Fox News poll indicating that 71% of Americans oppose local AI data centers.

Privacy Enhancements for ChatGPT Users

With increasing reliance on AI like ChatGPT for personal advice, privacy concerns have grown. OpenAI introduced Locked Chats, a new feature allowing users to protect individual ChatGPT conversations using a four-digit PIN, Face ID, Touch ID, or phone unlock methods.

Setting Up Locked Chats

To lock a conversation, sign into ChatGPT. Locate the chat you want to protect, open its menu, and select ‘Lock chat’. Confirm the action and create a PIN if prompted. Conversations move to Projects > Locked chats. They won’t appear in regular search results and need a PIN to unlock.

Accessing Locked Chats

Biometric authentication is supported on iPhones, while device authentication is available on some Android phones. To enable it, navigate to Security and login settings and activate Face ID or Touch ID for quick access to locked chats.

Opening and Managing Locked Chats

Access locked conversations under Projects. Enter your PIN or use biometric authentication to open them. You can continue chats without removing their protection but cannot archive, pin, move, or share locked conversations without unlocking them.

Privacy Limitations

It’s important to note that Locked Chats don’t offer end-to-end encryption nor alter how OpenAI retains information. Shared links remain active even if the conversation is locked. Organization accounts follow retention policies, and compliance tools might access locked chats.

Disabling Model Training

If desired, users can disable ‘Improve the model for everyone’ in their settings, preventing new conversations from assisting in OpenAI’s model training.

Comparison With Temporary Chats

Temporary Chats are for interactions you don’t want saved, differing from Locked Chats which allow you to revisit conversations. Choose according to the privacy desired for your interactions.

PIN Management and Recovery

Resetting a forgotten PIN deletes all locked chats permanently. If you wish to change your PIN, remember it to avoid losing conversations.

Conclusion

OpenAI’s Locked Chats improve control over who sees chat history but aren’t foolproof. Users should be cautious with the information they share, considering the current privacy limitations.

TAGS: