OpenAI has officially announced ChatGPT Computer History, a new capability for its macOS app that records your digital activity to help the AI assist you better. The feature is designed to turn your interactions across apps and websites into a searchable timeline, allowing ChatGPT and Codex to reference your recent work and suggest automations.
While OpenAI describes this as a way to help you pick up where you left off, the immediate reaction from the tech community has been a familiar one. Many users are drawing direct comparisons to Windows 11’s Recall feature, which faced significant backlash when it was first introduced. The core question on everyone’s mind is whether OpenAI has addressed the privacy concerns that plagued Microsoft’s approach.
Here is everything you need to know about what ChatGPT Computer History does, how it differs from Recall, and what the privacy implications mean for your Mac.
What is ChatGPT Computer History?
ChatGPT Computer History is an opt-in feature available for Pro, Business, and Enterprise subscribers. It is turned off by default, meaning your Mac will not start recording your activity the moment you update the app.
According to OpenAI’s documentation, the feature works by recording interaction events rather than capturing your screen. This means it tracks clicks, typing, app switching, and keyboard shortcuts. It does not record audio or take periodic screenshots of your display.

The data collected from these interactions is processed to create “memories” that ChatGPT can use. OpenAI states that this data is saved locally on your Mac, but it does get sent to OpenAI’s cloud servers for processing. Once the data is transformed into memories, it is sent back to your Mac for local storage.
OpenAI emphasizes that it does not retain this data on its servers unless required by law. Furthermore, the company confirms that it does not use this interaction data for training its models. This distinction is crucial for users concerned about their personal workflows being used to improve OpenAI’s broader AI systems.
How It Compares to Windows Recall
The similarities between ChatGPT Computer History and Windows 11’s Recall are hard to ignore. Both features aim to create a searchable history of your computer usage, allowing an AI assistant to understand your context and help you complete tasks.
However, the implementation differs significantly. Windows Recall relied on taking snapshots of your screen at regular intervals. This approach raised immediate privacy concerns, as it could inadvertently capture sensitive information like passwords, private messages, or financial data displayed on your monitor.
Microsoft delayed the launch of Recall by nearly a year to address these security issues. It eventually released in April 2025 as an opt-in feature for Copilot+ PCs. Even then, many users remained skeptical of the potential risks.

By choosing to log interaction events instead of capturing screenshots, OpenAI has taken a different path. This method is less invasive in terms of visual data, as it only records what you interact with rather than what is visible on your screen. It feels more like a structured log of your actions rather than a surveillance tool.
Despite this difference, the fundamental concept remains the same: an AI assistant is constantly monitoring your activity to build a profile of your work habits. For users who were wary of Recall, this similarity is likely to trigger similar hesitations.
Privacy and Security Considerations
While the shift from screenshots to interaction logging is a positive step, OpenAI’s documentation highlights several security caveats that users should be aware of.
First, the feature does not encrypt your information. This means that the data stored locally on your Mac is accessible to other apps running on the same system. If your Mac is compromised or if you have untrusted software installed, this data could potentially be exposed.

Second, there are risks associated with prompt injection. Because these AI-powered features can see what you are doing, they can be manipulated by text. If you visit a website containing hidden instructions designed to trick an AI, it could potentially perform unwanted actions based on your recent activity.
OpenAI notes that it does not keep the data on its servers unless required by law. However, the process of sending data to the cloud for processing inherently introduces a level of risk. Users should weigh the convenience of AI-assisted workflows against the potential exposure of their interaction data.
What This Means for You
For macOS users, ChatGPT Computer History represents a significant shift in how AI assistants can interact with your computer. The opt-in nature and the lack of screenshotting are clear improvements over previous attempts at continuous monitoring.
If you find value in having an AI that remembers your context and can suggest automations, this feature may be worth exploring. However, you should proceed with caution. The lack of encryption and the potential for prompt injection risks mean that this tool is not without its dangers.
It is advisable to review OpenAI’s privacy settings regularly and disable the feature if you notice any unusual behavior. As with any tool that monitors your activity, staying informed about how your data is handled is essential.
How to Get It
ChatGPT Computer History is available now for Pro, Business, and Enterprise subscribers on macOS. You can enable the feature within the ChatGPT app settings. If you are not a subscriber, you will need to upgrade your plan to access this functionality.
As with any new AI feature, Microsoft and OpenAI continue to navigate the fine line between helpful automation and user privacy. Your feedback and usage patterns will likely influence how these tools evolve in the coming months.
Source: Latest from Windows Central
Over to you: Are you turning on ChatGPT Computer History on your Mac, or is the privacy trade-off too steep for you?



