ChatGPT Gains an Opt-In Memory of Computer Activity
OpenAI’s Computer History lets ChatGPT recall recent activity across Mac apps and websites, with a 48-hour event-recording window and organization-level controls.
From conversation memory to computer memory
OpenAI has begun rolling out Computer History, an opt-in feature that allows ChatGPT to use a record of recent activity across applications and websites on a Mac. Instead of relying only on chats, uploaded files and connected services, the assistant can draw on a broader timeline of what the user has recently been doing and resume work with less repeated explanation.
The feature is launching in the ChatGPT desktop application for Mac for Pro, Business and Enterprise customers. OpenAI says availability is rolling out globally, although access in the European Economic Area, the United Kingdom and Switzerland will follow later. Computer History is off by default. Pro users can enable it in the app’s Computer History settings, while Business and Enterprise users require their organization’s administrator to make it available.
Computer History builds on OpenAI’s earlier Chronicle research preview. Rather than taking screenshots or recording audio, it records interaction events from selected apps and websites. OpenAI says those raw events are stored temporarily for no more than 48 hours while they are processed into memories. Users can inspect their timeline, remove individual activity and clear associated memories. OpenAI has not presented the feature as an unlimited archive of raw computer activity; its purpose is to supply recent working context to future ChatGPT and Codex interactions.
The distinction matters because a chronological record of computer use can contain far more sensitive information than an ordinary chat history: documents opened, sites visited and fragments of work spanning multiple applications. An opt-in design, visible timeline, short raw-event retention period and deletion controls reduce some risk, but organizations will still need to decide whether the productivity gain justifies exposing this category of context to an assistant. The delayed European rollout also indicates that regional privacy and compliance requirements remain operational constraints.
Why it matters
Computer History moves the AI assistant closer to being a persistent layer over the operating system. If it works reliably, users may no longer need to manually assemble context before delegating a task, removing one of the largest sources of friction in agent workflows. At the same time, it makes control over collection, retention and deletion a core product battleground. The companies that win personal AI may be those trusted to remember the most—not merely those with the strongest underlying model.