A new feature in the ChatGPT app for macOS is designed to give the AI assistant more context about how someone uses their computer. It reflects a wider shift as AI assistants gain greater awareness of users’ activity and work across their devices, with related capabilities appearing in services from Microsoft, Anthropic, and others. Kaspersky says the privacy and security questions this creates will become increasingly relevant across operating systems as these tools become more deeply integrated into everyday computer use.
Computer History arrived in the ChatGPT app for Mac in mid-August and is available to Pro users and above. It is disabled by default, so anyone who wants to use it must enable it and grant the necessary macOS permissions.
Once activated, ChatGPT uses macOS accessibility features to capture information such as text visible on screen, window titles, text being typed into fields, and interactions with applications. It does not record audio, read masked password fields, or view the contents of images and videos. OpenAI also says private browsing windows are excluded.
The raw activity data stays on the Mac for up to 48 hours. During that time, ChatGPT periodically creates summaries of related activity, effectively turning recent computer use into diary-style entries. Kaspersky says those summaries are saved as plain text files in a folder that other applications on the Mac can read. When these summaries are later pulled into ChatGPT conversations, the usual ChatGPT privacy settings apply.
“The appeal is easy to understand because an AI assistant that knows what you have been working on can respond with much more useful context,” says Brandon Muller, Senior Security Consultant for the META region at Kaspersky. “The trade-off is that information from across your day is being brought together and summarised in one place. That can become valuable to somebody who should not have access to it.”
Kaspersky points in particular to the risk from macOS infostealers. Malware of this kind is designed to look for useful data stored locally on a device, and attackers tend to adapt quickly when new sources of information become available. A Computer History summary may not contain a password or card number, but it could still reveal enough about someone’s work, communications or recent activity to support a convincing targeted scam.
The concern is a familiar one. Microsoft’s Recall feature also aims to make previous computer activity easier to find, but it works differently by capturing screenshots and analysing their contents. Microsoft strengthened Recall considerably between its original announcement and its 2025 release, including encryption and user-controlled activation, but Kaspersky continued to recommend caution for people who routinely handle sensitive information.
For Mac users who plan to use Computer History, Kaspersky recommends deciding upfront what ChatGPT should be allowed to track. Apps and websites containing private conversations, financial information, or confidential work should be manually added to the exclusions list, while the history itself should be reviewed regularly to ensure that sensitive sources have not been included unintentionally. The Mac should also be protected against unauthorised access and malware via a suitable security solution such as Kaspersky Premium for macOS.
People whose work involves confidential client, medical or other protected information may be better served by severely limiting Computer History or leaving it disabled.
“Features like this show where AI assistants are heading. The more context they have, the more useful they can become, but that context also deserves protection. Users should understand what is being collected, exclude information they would not want exposed, and make the choice based on the kind of work they do on that machine,” says Muller.
