Computer History in One Sentence: Opt-In, Always-On Activity Tracking for ChatGPT
Computer History is an opt-in ChatGPT activity tracking feature in the Mac app that records your clicks, keystrokes, app switches, and website interactions as an “interaction-event stream,” turns them into searchable memories and timelines, and then sends those events to OpenAI’s servers to improve context-aware responses. That’s the trade: you give the AI a detailed shadow of your daily computer life in exchange for smarter help with your work. OpenAI positions this as a friendlier alternative to screenshot-based monitoring, but it is still, in plain language, “a keylogging and event capture system.” If you switch it on, you are not enabling a minor convenience toggle—you are agreeing that your Mac app data collection becomes fuel for an AI that lives beyond your device.
From Screenshots to Keylogging: A Softer Story, Same Hard Privacy Cost
OpenAI is careful to frame Computer History as a privacy upgrade over its Chronicle experiment, which captured screenshots of your screen activity. Computer History avoids images, microphone input, and system audio, instead building that interaction-event stream from allowed apps and websites—“events can include clicks, typing, keyboard shortcuts, app switches, and context that macOS exposes through its accessibility system.” That sounds less intrusive than full-screen grabs, but it does not change what this is: continuous Mac app data collection specifically designed to follow what you do so ChatGPT and Codex can use it later. According to OpenAI’s own documentation, Computer History files are stored unencrypted locally for up to 48 hours and then processed on OpenAI’s servers to generate memories. The result is a powerful AI privacy tradeoff: more meaningful, context-rich answers in exchange for a detailed log of how you work.
Who Is Affected—and How Much Control Do They Really Have?
Computer History is off by default and only available in the ChatGPT desktop app on macOS for Pro, Business, and Enterprise users. It is not offered through API access or other hosting routes. Pro users can turn it on themselves, while Business and Enterprise members need an admin to explicitly grant access before any individual can opt in. There are controls: you decide whether to enable Memories, which apps and websites are allowed, and you can clear recorded activities so ChatGPT stops using them. But control has limits. Once you opt in, those activity events leave your Mac for processing, even though OpenAI says it does not retain raw event files after they are turned into memories unless required by law. Those generated memories then return as plain-text Markdown files on your machine—files OpenAI warns “can contain sensitive information” and “are not encrypted by Computer History.” Other programs running as your macOS user may be able to read them.
The Governance Problem: Workplace Keylogging by Consent
For businesses, Computer History is not a quirky productivity tool; it is a new category of workplace surveillance your staff can opt into. Once enabled, it captures clicks and keystrokes as a continuous stream, and those events are processed by an external AI system that can reuse the resulting memories in future chats. The business issue is clear: this click-and-keystroke activity stream is “a new category of workplace context collection, even without screen images.” It also expands the prompt injection attack surface, because more of an employee’s context is available to hostile content if someone pastes a shady script or visits a malicious page while Computer History runs. Governance needs to catch up. The sensible advice for leaders right now is explicit: do not roll this out across your workspace yet, and do not ignore it either. Confirm your Business or Enterprise admin console still has Computer History turned off, decide who is allowed to change that setting, and define written rules before any pilot starts.
What You Should Do Today: Treat It as Surveillance, Not a Toy
If you are an individual user, the core question is not “will this make ChatGPT more helpful?”—it will—but “am I comfortable with a friendly keylogger shaping my AI experience?” Computer History seems like a useful way to incorporate your activities into ChatGPT conversations and to organize your work, yet it “sounds like a potential privacy risk.” The practical stance is cautious experimentation, not blind adoption. Turn it on only if you understand that your activity timeline is being processed off-device, that memories may contain sensitive information, and that those memory files are stored in plain text other apps can see. At minimum, consider pausing it or excluding apps that contain health, financial, or personal data. For companies, the bar is higher: do not roll out Computer History widely, keep the admin switch off, and authorize only a small test group operating under written governance, with regular audits of what is being captured. The real productivity upgrade here is not the feature—it is learning to say no until the privacy costs are worth paying.
