ChatGPT gets more personal by default
OpenAI says the model used by hundreds of millions of people has fewer hallucinations.
TL;DR
- OpenAI updated ChatGPT's default model to GPT-5.5 Instant, featuring more accuracy and personalization.
- GPT-5.5 Instant reduced hallucinated claims by 52.5% and inaccurate claims by 37.3% in testing.
- Enhanced personalization, using past chats and connected services, is rolling out first to Plus and Pro users.
- New 'memory sources' controls allow users to view, delete, or correct ChatGPT's personalized context.
- The increased reliance on personal data for chatbot responses raises privacy concerns, despite new controls.