ChatGPT’s Computer History tracks your clicks and keystrokes

OpenAI has introduced Computer History to its macOS ChatGPT desktop app, an opt-in feature that monitors user clicks and keystrokes to build a personal activity timeline. The data helps ChatGPT and Codex suggest workflow automations and resume unfinished tasks, while private browsing sessions are automatically excluded.

OpenAI has rolled out a feature called Computer History for its macOS ChatGPT desktop app that records user clicks and keystrokes to construct a personal activity timeline. This log can be referenced by both ChatGPT and Codex to better understand how a user works, propose automation shortcuts, and even pick up tasks the user left incomplete. Unlike many data-collection features that require users to actively opt out, Computer History is opt-in by default, giving users a clearer choice before any monitoring begins. Users retain additional control by blacklisting specific apps or websites from tracking, and individual timeline entries can be manually deleted. OpenAI's product engineering lead confirmed that activity occurring within incognito or private browser tabs is automatically excluded. The feature represents OpenAI's push to make ChatGPT a more persistent, context-aware assistant embedded into daily computing workflows rather than a standalone chatbot.

OpenAI has launched a feature called Computer History inside its macOS ChatGPT desktop application, marking a significant step toward turning the AI chatbot into a persistent, always-aware productivity assistant. The feature tracks mouse clicks and keystrokes to assemble a chronological log of what a user does on their computer. That timeline then becomes a reference point for both ChatGPT and OpenAI's coding-focused model Codex, enabling them to tailor suggestions, propose workflow automations, and even resume tasks a user abandoned partway through.

On the privacy front, OpenAI has made the feature opt-in rather than opt-out — a meaningful design choice that signals at least some awareness of user concerns around continuous screen monitoring. Users also have granular controls: specific applications or websites can be excluded from logging, and individual timeline entries can be deleted on demand. Activity inside private or incognito browser windows is automatically ignored, according to an OpenAI engineering manager.

Why it matters: Computer History is part of a broader industry race to build AI agents that understand context over time rather than responding only to one-off prompts. By embedding itself into daily computing habits, ChatGPT is positioning itself less like a search engine and more like a long-term digital collaborator that knows your work patterns. That is compelling from a productivity standpoint — an AI that remembers you left a report half-finished and offers to continue it is genuinely useful.

However, the implications deserve scrutiny. Keystroke and click logging is the same class of data that security researchers classify as highly sensitive, since it can reveal passwords, private communications, and browsing habits. Even with opt-in design and deletion options, questions remain about how long this data is stored, whether it is used to train future models, and how it might be handled in the event of a data breach or legal request. As AI assistants grow more capable, the trade-off between convenience and surveillance-style data collection will increasingly define how users and regulators respond to these tools.