ChatGPT Can Now Remember What You Do on Your Mac, And That Raises a Big Privacy Question
ChatGPT Can Now Remember What You Do on Your Mac, And That Raises a Big Privacy Question
What if ChatGPT did not just remember what you told it, but could also remember what you were doing on your computer?
That is the idea behind OpenAI's new Computer History feature for the ChatGPT macOS app.
And yes, it sounds a little like Microsoft's controversial Windows Recall. But there is an important difference: OpenAI says its system does not take screenshots, record video, or capture audio. Instead, it records computer activity as "events" to build a timeline of what you have been doing.
That sounds less dramatic.
Until you think about what an AI could do with that information.
Your AI Could Know What You Were Working On
Imagine spending the morning jumping between Google Docs, Slack, your browser, and other apps.
Later, you ask ChatGPT:
"What was I working on this morning?"
Instead of starting from zero, Computer History can use its timeline to provide context.
OpenAI's demonstration showed the feature retrieving a recently edited document, checking whether it had been shared with colleagues, and summarizing the user's morning activity.
The bigger idea is that ChatGPT becomes less like a chatbot you visit and more like an AI assistant that understands your digital workspace.
It could potentially help you pick up unfinished tasks, identify patterns in your workflow, and suggest ways to automate repetitive work.
But No, ChatGPT Is Not Secretly Recording Your Screen
This is where the headline can be misleading.
Computer History is not designed like Windows Recall, which became controversial because of its screenshot-based approach.
OpenAI says Computer History works with events rather than visual captures. That means it is not supposed to store a continuous collection of screenshots showing everything on your display.
The feature is also opt-in, meaning users have to choose to enable it.
Users can control which applications and websites are included, exclude specific apps or sites, and delete individual history entries. OpenAI's Ari Weinstein has also said activity from incognito or private browser sessions is automatically excluded.
That gives users considerably more control than the original concerns surrounding Microsoft's Recall.
So, What Is the Catch?
Even without screenshots, the idea of an AI keeping a detailed record of your computer activity raises obvious privacy questions.
Think about how much information passes through your computer every day.
Emails.
Work documents.
Private conversations.
Research.
Financial websites.
Business dashboards.
Social media.
The fact that an AI does not need a screenshot to understand parts of your workflow is actually what makes this technology so interesting and potentially sensitive.
And there is another important correction to the original claim: Computer History is not simply "turning everything you do into training data."
Recent reporting says the feature's activity history is used to provide contextual assistance rather than automatically becoming training material for OpenAI's models.
That distinction matters.
OpenAI Is Building a More Persistent AI
The timing is also significant.
OpenAI has been steadily turning ChatGPT's desktop experience into something that can actually work across your computer, rather than answer questions.
Its newer desktop experience combines ChatGPT, Work, and Codex, while Work can interact with local files and desktop applications with permission.
Computer History takes that idea one step further.
Instead of repeatedly explaining what you were doing, the AI can look at your recent workflow and understand the context on its own.
That is powerful.
But it also creates a new question for the AI era:
How much should your AI know about you before it becomes too much?
Computer History suggests OpenAI believes users want an assistant that remembers more of their digital lives.
Whether users are comfortable giving it that much context may be the real test.