Skip to content
ViqusViqus
Navigate
Company
Blog
About Us
Contact
System Status
Enter Viqus Hub

ChatGPT Adds PIN-Protected 'Locked Chats' for Enhanced Local Privacy

ChatGPT Privacy PIN Protection Local Security Language Models OpenAI
October 20, 2026
Source: TechCrunch AI

This summary and analysis were generated by AI from the original article at TechCrunch AI and may contain errors (how Viqus works). Read the source for full details.

Viqus Verdict Logo Viqus Verdict Logo 3
Minor UX Polish, Not a Privacy Paradigm Shift
Media Hype 4/10
Real Impact 3/10

Article Summary

OpenAI has rolled out a new 'Locked Chats' feature for ChatGPT, enabling users to secure sensitive conversations with a four-digit PIN or biometric authentication. This feature is designed for temporary privacy when sharing a device, distinguishing it from Temporary Chats. While it isolates locked chats into a separate section, OpenAI explicitly clarifies that this does not equate to end-to-end encryption or change data retention policies. The functionality is available across all current ChatGPT plans, though it excludes specific chat types like group chats and those from shared links. This addition signals an increased focus on user-perceived privacy controls for personal use cases.

Key Points

  • Users can now secure specific ChatGPT conversations using a PIN or biometrics via the new 'Locked Chats' feature.
  • This feature adds a layer of local privacy but does not constitute end-to-end encryption or alter core data retention policies.
  • Locked Chats are available across nearly all ChatGPT plans, though certain chat types remain excluded from the protection.

Why It Matters

This is a routine, incremental feature release focused on user experience and perceived privacy rather than a fundamental shift in the underlying AI capabilities or data architecture. While adding a visible control layer, the detailed caveats—such as the lack of true end-to-end encryption—mean the industry impact is negligible. It addresses user anxiety around device sharing but does not change how the model processes or stores data fundamentally.


You might also be interested in