Skip to main content

ChatGPT for Mac Now Tracks Your Work History

OpenAI has quietly rolled out a new feature for the ChatGPT desktop app on macOS that might make you feel like the AI is actually paying attention to what you do all day. Called "Computer History," it tracks your activities on the computer and turns them into a timeline that ChatGPT and its coding companion, Codex, can reference. The idea is to give the AI a better sense of how you work, so it can offer suggestions, pick up where you left off, and even help with tasks you haven't finished yet.

This isn't a Big Brother-style surveillance tool, though. The feature is opt-in, meaning you have to switch it on before it starts recording anything. And if you're worried about privacy, you can exclude specific apps and websites from being tracked, and you can delete any recorded history at any time. OpenAI's product and engineering manager, Ali Weinstein, also mentioned that the system automatically ignores anything you do in incognito or private browser tabs, so your secret late-night Wikipedia rabbit holes stay off the record.

In a demo, the feature showed how it could help you find a document you edited recently, confirm whether you shared a file over Slack, or give you a quick summary of what you worked on that morning. It's a bit like having a personal assistant who's been watching over your shoulder, but in a helpful, non-creepy way.

What sets this apart from Microsoft's Windows Recall, which also aims to remember your digital footsteps, is that ChatGPT's version doesn't rely on screenshots. OpenAI is clear that it doesn't capture images, videos, or audio. Instead, it logs your actions as "events"—like opening a file, sending a message, or visiting a website—and builds a timeline from those. This approach is less intrusive and arguably more respectful of your privacy, since it's not storing visual snapshots of your screen.

From a product standpoint, Computer History is a big step toward making ChatGPT more proactive. Instead of just answering questions when you ask, it can now sense what you're working on and offer help before you even realize you need it. For example, if you've been editing a report all morning, ChatGPT might suggest a better way to phrase a paragraph or remind you to save your work. It's a shift from a reactive tool to a collaborative partner.

Of course, with any feature that tracks your activity, there are bound to be questions about data security and user trust. OpenAI seems to be addressing those by putting control in the user's hands. You decide whether to enable it, what gets recorded, and when to wipe the slate clean. The company also emphasizes that the data stays within the app and is used to improve your experience, not for advertising or anything sketchy.

As AI becomes more integrated into our daily workflows, features like this are likely to become the norm. The challenge for companies like OpenAI is to balance usefulness with privacy, and so far, this opt-in, event-based approach seems like a reasonable middle ground. Whether users will embrace it or find it too nosy remains to be seen, but it's definitely a sign of where AI is heading—toward a more intuitive, context-aware assistant that's there for you, even when you're not talking to it.

Key Points

  • Opt-in and controllable: Users can enable or disable Computer History, exclude specific apps, and delete history anytime.
  • Event-based, not screenshot-based: Unlike Windows Recall, it logs actions as events, not images or audio.
  • Proactive assistance: The feature helps ChatGPT understand your workflow and offer suggestions or complete tasks.
  • Privacy safeguards: Incognito tabs are ignored, and data is used to improve user experience.
  • Product evolution: Marks a shift from reactive AI to proactive, context-aware assistance.