I'm excited about ChatGPT's memory upgrade - but I'm quickly seeing a downside

2026-06-18 · Source: News and Advice on the World's Latest Innovations | ZDNET · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy · Depth: Intermediate, long

Summary

ChatGPT has introduced significant memory upgrades, evolving from a simple list of facts in 2024 to "Dreaming V3" in 2026, which synthesizes a user profile from entire chat histories, explicit instructions, and implicit preferences. While OpenAI reports improved factual task recall from 41% (2024) to 82% (2026), preference adherence from 31% to 71%, and a 5X compute cost reduction, the feature raises concerns. The system often retains outdated or irrelevant information, leading to inaccurate responses, such as incorrectly stating user experience with Kasa smart plugs or Assistant. Users can partially disable memory or delete saved facts, but chat history remains, and safety-relevant context persists. This profiling capability, available to Plus and Pro subscribers and rolling out to all users, can skew responses based on a potentially flawed internal representation of the user.

Key takeaway

For AI product managers or professionals relying on ChatGPT for research, be aware that its "Dreaming V3" memory feature can create skewed, inaccurate user profiles. You should actively review and correct your consolidated profile via the "Manage" button in settings. Understand that disabling memory or deleting saved facts does not fully purge your chat history, and safety-relevant context persists. This necessitates increased vigilance in verifying AI responses, as they may be filtered through a flawed, persistent personal lens.

Key insights

ChatGPT's advanced memory, "Dreaming V3," creates persistent user profiles, but risks inaccuracy and privacy concerns.

Principles

• AI memory systems can misinterpret user context.
• User-curated memory often goes stale quickly.
• Disabling AI memory may not fully erase data.

Method

To manage ChatGPT's memory: access Settings > Personalization > Memory, then disable "Enable memory" (partially), delete saved memories, or use "Manage" to edit the consolidated profile.

In practice

• Review ChatGPT's consolidated profile.
• Delete irrelevant saved memories.
• Delete entire chats for full data purge.

Topics

• ChatGPT
• AI Memory
• User Profiling
• Data Privacy
• AI Accuracy
• OpenAI

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Product Manager, Tech Journalist, Consultant

Related on AIssential

• ChatGPT's new memory upgrade is powerful - and could poison every answer it gives you — ChatGPT's advanced memory system, while technically efficient, creates an often inaccurate and unmanageable user profile that can bias responses.
• 😺 ChatGPT admitted its memory was broken — OpenAI's memory upgrade for ChatGPT significantly improves recall and adherence, revealing past accuracy issues in AI systems.
• ChatGPT Memory Got a Much Needed Update! — ChatGPT's upgraded memory system now allows users to directly manage and correct stored information, improving conversational accuracy.
• Dreaming: Better memory for a more helpful ChatGPT — ChatGPT's "dreaming" memory system automatically curates and updates user context for personalized, current, and relevant interactions.
• How to audit what ChatGPT knows about you - and reclaim your data privacy — Users should proactively manage personal data shared with ChatGPT due to uncertain future usage.
• I'm a ChatGPT power user: Here are 7 useful settings that are turned off by default — ChatGPT offers extensive customization options to tailor appearance, model behavior, personality, and privacy settings.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by News and Advice on the World's Latest Innovations | ZDNET.