OpenAI may be testing Windows Recall-like feature for its Codex coding agent: What it means for users

ChatGPT maker OpenAI is reportedly testing a new feature for its AI coding agency, Codex. This feature, called “Chronicle,” has drawn comparisons to Microsoft’s controversial Windows Recall feature. OpenAI’s new opt-in research preview allows Codex to capture screenshots of a user’s screen and use that visual information to build contextual “memories,” helping the AI agent understand ongoing work without requiring users to repeatedly explain what they are doing. The feature is currently available only in the Codex app for macOS and has already raised privacy and cybersecurity concerns.According to OpenAI’s documentation seen by The Register, Chronicle is designed to improve Codex’s contextual understanding by capturing what users are working on. In the document, OpenAI noted, “Chronicle augments Codex memories with context from your screen. When you prompt Codex, those memories can help it understand what you've been working on with less need for you to restate context.”The feature periodically captures screenshots of a user’s screen and processes selected screenshots using OCR and other tools to extract text and context. That information is then converted into text-based memory files that Codex can reference in future sessions.The feature has been compared to Windows Recall, which Microsoft rolled out in 2024. Recall automatically takes screenshots of your desktop every few seconds and saves them locally to make the Copilot with AI assistance even better. Privacy researchers have criticised the feature, warning that screenshots could be used to store sensitive information.Security researcher Michael Taggart also highlighted similarities between the two systems and wrote: “Oh my god, OpenAI reinvented Recall, but for macOS.” However, unlike Recall, Chronicle is currently optional and available only on macOS.OpenAI said screenshots are temporarily stored on-device for up to six hours. The company noted that selected screenshots are sent to OpenAI servers for processing to generate text-based memories.According to documentation, those screenshots are not stored permanently or used for training. However, the generated memory files remain stored locally until users manually delete them. These files are reportedly saved in: $CODEX_HOME/memories_extensions/chronicle/OpenAI said users should be aware of potential risks. The company wrote, “Both directories for your screen captures and memories might contain sensitive information. Make sure you do not share content with others, and be aware that other programs on your computer can also access these files.”OpenAI has also acknowledged that Chronicle may create new risks. “Before enabling, be aware that Chronicle uses rate limits quickly, increases risk of prompt injection, and stores memories unencrypted on your device,” the company notedThis means users may consume Codex limits faster while also exposing themselves to prompt-injection attacks if malicious content appears on-screen and is captured by the AI. The feature currently requires users to manually enable it through Codex settings and grant screen recording permissions. It is available as a research preview for eligible macOS users and is not currently available in the EU, the UK, or Switzerland.For developers, Chronicle could reduce repetitive prompts by helping Codex automatically understand coding errors, files, and workflows. However, users handling sensitive information may need to carefully review privacy implications before enabling the feature, especially since screen captures and stored memory files may contain passwords, private messages, documents or financial data.