OpenAI has launched a new capability for its ChatGPT desktop application on Mac that builds a searchable timeline of a user’s activity across apps and websites, a move aimed at boosting productivity but likely to draw scrutiny from privacy-conscious users.
The feature, called Computer History, began rolling out globally on August 13 for Pro, Business, and Enterprise subscribers. Users in the European Economic Area, the United Kingdom, and Switzerland will gain access in the coming weeks. The tool is opt-in, meaning no data is collected unless a user explicitly enables it through the app’s Settings menu under Integrations.
Computer History replaces the Chronicle research preview, which OpenAI had previously introduced to monitor screen activity. The new version consumes fewer tokens and offers more granular privacy controls, according to OpenAI. Unlike Chronicle, which captured screenshots, Computer History records interactions with apps and websites but does not capture the screen, microphone input, or system audio.
The feature creates what OpenAI describes as an “interaction-event stream” from permitted apps and websites. Events can include clicks, typing, keyboard shortcuts, app switches, and context exposed through MacOS’s accessibility system. The system periodically converts these events into text summaries and local memory files, which are then organized into a timeline by day and time.
Each captured item in the timeline displays a title, a summary of the activity, the apps or websites used, a suggested skill for repeatable work, and options to view the memory file in Finder or delete the item entirely.
OpenAI has positioned the tool as a way to reduce the amount of context users need to type manually. The company provided several example queries users could direct at ChatGPT based on the timeline, including “What was I working on before my last break?” and “Give me a list of tasks I’ve worked on today and their status.” The feature also integrates with Codex, OpenAI’s coding tool, allowing developers to pick up where they left off on projects.
To use Computer History, users must also enable the Memories feature, since the tool relies on it to apply activity context across chats and tasks. Users can select which apps and websites to include, pause collection at any time, and clear recorded history either entirely or for specific time periods.
Despite these controls, OpenAI has acknowledged a notable limitation. In its documentation for the feature, the company states that Computer History files can contain sensitive information and are not encrypted. Other programs running under the same MacOS user account may be able to access them. OpenAI advises users to protect their Mac accounts and exclude sources they do not want included.
That lack of encryption is likely to be a sticking point for security-minded users. While the feature does not transmit data to OpenAI by default beyond what is used in conversations, locally stored memory files could be vulnerable to malicious software or unauthorized access on a shared machine.
The launch comes amid a dense product cycle for OpenAI. Hours before the Computer History rollout, the company announced GPT-5.6 Ultrafast mode, powered by Cerebras hardware, which generates up to 750 tokens per second for API customers. The rapid pace of releases reflects OpenAI’s broader push in 2026 to embed persistent memory features across its product line.
Privacy regulators in Europe have kept a close watch on such tools. The decision to delay the EEA, UK, and Switzerland rollout suggests OpenAI is navigating regulatory requirements before expanding access in those markets.
For now, Computer History remains a Mac-only feature, with no indication from OpenAI about when or whether it will arrive on Windows or mobile platforms. Users interested in testing the feature should weigh the productivity benefits against the security implications of storing unencrypted activity logs on their machines.
Note: Ziff Davis, the parent company of ZDNET, filed a lawsuit against OpenAI in April 2025 alleging copyright infringement in the training and operation of its AI systems.