ChatGPT Computer History: AI Tracks Your Clicks | Guide

What happened

OpenAI just rolled out ChatGPT Computer History, a new feature inside the ChatGPT desktop app for macOS that quietly logs your clicks, keystrokes, and app activity to build a searchable timeline of your workday. Spotted on August 16, 2026, the feature turns everyday actions — opening a document, switching between apps, sending a Slack message — into structured "events" that ChatGPT and Codex can reference the next time you ask for help.

Unlike a simple activity log, Computer History is designed to learn how you work. It can suggest automations based on repeated patterns, and — in one demo shared by OpenAI's Developer Experience team — it picked up a half-finished task, checked whether a document had been shared via Slack, and delivered a recap of the user's entire morning without being asked for specifics.

The feature is opt-in, not opt-on-by-default, and OpenAI has built in several controls: you can exclude specific apps or websites from being tracked, and you can delete individual entries after the fact. According to Ari Weinstein, OpenAI's Product and Engineering manager, Computer History automatically ignores anything that happens inside incognito or private browser tabs.

Why it matters

Computer History arrives less than two years after Microsoft's Windows Recall feature triggered a wave of privacy backlash for capturing constant screenshots of user activity. OpenAI is clearly trying to avoid that same reaction by taking a different technical approach: instead of storing images, video, or audio, Computer History logs discrete text-based "events." That's a meaningful distinction — a screenshot can capture passwords, private messages, or financial data verbatim, while an event log can be scoped more narrowly to what app was used and what action was taken.

Still, the underlying goal is nearly identical to Recall's: give an AI assistant persistent memory of your digital behavior so it can act more like a personal chief of staff than a chatbot that forgets everything between sessions. For a company racing to make ChatGPT indispensable in daily workflows — especially through Codex, its coding assistant — building this kind of continuous context is a strategic necessity, not a side feature.

This also signals where AI assistants are heading industry-wide. Google, Microsoft, and OpenAI are all converging on the same idea: an AI that observes your computer use passively, rather than one that only responds when prompted. Expect competitors to ship similar "memory of your machine" features within the next year.

How to use it today

Computer History is currently limited to the ChatGPT desktop app on macOS, and it has to be turned on manually in Settings — it does not activate automatically. Once enabled, ChatGPT and Codex can pull from your recent activity timeline whenever you ask a question that references past work, such as "what was I working on this morning" or "pick up where I left off on that report."

Before turning it on, it's worth doing a quick audit of which apps and sites you actually want included. OpenAI lets you exclude specific apps or websites from tracking, so consider blocking anything tied to banking, healthcare, HR systems, or other sensitive tools before enabling the feature broadly.

If you're experimenting with AI-assisted workflows more broadly — automations, quick prototypes, or lightweight productivity tools that don't require this level of activity tracking — a set of [free AI tools at mykreatool.com](https://mykreatool.com) can be a lower-commitment starting point while you decide how comfortable you are with a feature like Computer History watching your screen all day.

Who benefits

Developers and Codex users stand to gain the most immediately. Because Computer History feeds directly into Codex's context, it can help the coding assistant understand what file you last edited, what documentation you were reading, or which terminal commands preceded a bug — cutting down on the repetitive back-and-forth of re-explaining context in every session.

Knowledge workers juggling many small tasks across apps — freelancers, consultants, marketers managing multiple client projects — could also benefit from the automatic recap function. Instead of manually reconstructing "what did I do this morning," the assistant can generate that summary on demand, which is useful for time tracking, status updates, or simply resuming focus after a context switch.

Small teams that already rely heavily on ChatGPT for daily operations may find the automation suggestions the most valuable part: if Computer History notices you repeatedly copy data from one app to another, it can flag that pattern as a candidate for a scripted automation, potentially saving hours per week.

Risks

The obvious concern is surveillance, even in opt-in form. Any system that logs keystrokes and clicks — regardless of how the data is stored — creates a detailed behavioral record of a person's work habits, and that record becomes a target for misuse if account security is compromised, if OpenAI's data-handling policies change, or if the feature is later enabled by default in a future update.

There's also an organizational risk for teams: employees using Computer History on work devices could inadvertently expose proprietary information, client communications, or credentials typed into non-excluded apps. Companies with strict data-governance policies will likely need to explicitly ban or restrict the feature until enterprise-grade controls are available.

Finally, the comparison to Windows Recall is a reminder that public trust in this category of feature is fragile. Microsoft ultimately delayed and reworked Recall after severe criticism from security researchers. OpenAI's screenshot-free approach is a genuine technical improvement, but "less creepy than Recall" is a low bar, and scrutiny from privacy advocates is all but guaranteed as adoption grows.

Conclusion

ChatGPT's Computer History is a meaningful step toward AI assistants that understand your work continuously rather than one prompt at a time, and its event-based, screenshot-free design is a real improvement over Microsoft's earlier misstep with Recall. For developers, Codex users, and anyone managing scattered daily tasks, the productivity upside is tangible. But the feature still asks users to trust a company with a granular log of their clicks and keystrokes, so anyone enabling it should start narrow — excluding sensitive apps, reviewing entries regularly, and treating the opt-in toggle as a decision worth revisiting, not a one-time setup step.