Ce prompt révèle tout ce que ChatGPT sait sur vous (flippant)

Ce prompt révèle tout ce que ChatGPT sait sur vous (flippant)

by Ludo Salenne

Applied Sciences & Engineering Computing & Cybersecurity URComputer securityURDPrivacy and data protection
🎙 Ludo Salenne 👥 266K 📅 October 3, 2025 ⏱ 22 min 👁 25K 📄 tutorial 🧭 2026-08-03
Available in: English (current) Français

Keywords

ChatGPT memoryhidden memorypromptprivacydata

Summary

In this video, Ludo Salenne explains the two types of memory in ChatGPT: visible (saved memory) and hidden (reference chat history). He demonstrates how to access the visible memory through settings and then introduces a custom prompt to extract the hidden memory, which contains information ChatGPT has learned from past conversations without explicit user consent. The prompt instructs ChatGPT to list all remembered information, including sensitive data, and to provide a JSON export. The video shows examples of the information retrieved, such as personal details, preferences, and behavioral patterns. It also discusses the implications for privacy, noting that OpenAI may use this data for training. The creator provides steps to disable the hidden memory and offers his opinion on the matter. The video includes a promotional segment for his paid course on automating ChatGPT.

134 words

Critical Evaluation

The video provides a practical tutorial on accessing ChatGPT’s hidden memory, which is a valuable and timely topic given growing concerns about AI privacy. The creator clearly explains the difference between saved memory and reference chat history, and he demonstrates a specific prompt that can be used to extract hidden data. The approach is methodical and includes safeguards such as requesting a JSON export and flagging sensitive information. However, the video has several limitations. First, it is heavily promotional, with a significant segment dedicated to selling a paid course, which may bias the content. Second, the claims about ChatGPT’s hidden memory capabilities are based on the creator’s personal experience and are not independently verified. While he references OpenAI’s official documentation, he does not provide direct evidence or citations for the specific behaviors he demonstrates. Third, the video lacks a critical analysis of the ethical and legal implications of accessing hidden memory, instead focusing on the technical ‘how-to’. The title uses sensational language (‘flippant’) which may overstate the shock value, but the content does reveal potentially surprising information. Overall, the video is useful for users wanting to understand and manage their ChatGPT memory, but it should be viewed with caution due to its promotional nature and lack of rigorous sourcing.

209 words

Title / Content Match

The title accurately reflects the content, which reveals a method to access hidden memory, though it uses sensational language.

Quality & Reliability

6/10

The video is a practical tutorial demonstrating how to access ChatGPT's hidden memory via a specific prompt. It is based on the creator's personal experience and references OpenAI's official documentation. However, it includes promotional content for a paid course and lacks independent verification of claims.

Chapters

Cited Sources

  • Memory FAQ - OpenAI Help Center — Official documentation explaining the two types of memory in ChatGPT: saved memory and reference chat history.
  • Trustpilot Reviews for SLN Web — Customer reviews for the creator's company, mentioned as social proof.
  • Formation ChatGPT (paid course) — Promotional link to the creator's paid course on automating ChatGPT.
  • Ressources IA gratuites — Free AI resources including prompts and mini-courses.

Concurring Sources

  • OpenAI Memory FAQ — Confirms the existence of two memory types and how they work.

External References

Contribution & Novelties

The video offers a practical prompt to access ChatGPT’s hidden memory, which is not widely known. It provides a step-by-step guide and highlights the privacy implications. The novelty lies in the specific prompt structure and the emphasis on behavioral tracking.

Pour aller plus loin :

76 words

Radar Profile

The radar profile shows moderate scores across all dimensions, indicating a balanced but not exceptional video. The highest score is in quantity of information, while technical level is lower, reflecting the tutorial nature.

Reliability 6/10