OpenAI Replaces Chronicle with Opt-In Computer History on Mac

OpenAI launched Computer History on August 13 for ChatGPT Pro, Business, and Enterprise users on macOS. The opt-in feature replaces the Chronicle preview with an interaction-event system that can turn activity in approved apps and websites into a timeline and local memories for ChatGPT and Codex, without capturing screenshots or audio. Temporary event files are retained for up to 48 hours.
OpenAI launched Computer History on August 13 as an opt-in feature in the ChatGPT desktop app for macOS. It turns activity from apps and websites a user permits into a timeline and local memories that ChatGPT and Codex can reference. The release is available to ChatGPT Pro, Business, and Enterprise users, although Business and Enterprise administrators must first enable access. Initial availability excludes the European Economic Area, Switzerland, and the United Kingdom.
A rebuilt replacement for Chronicle
Computer History replaces OpenAI's earlier Chronicle research preview, but the company describes it as a rebuilt system rather than a renamed version. Chronicle used screenshots. Computer History instead records interaction events exposed through macOS accessibility services, such as clicks, typing, keyboard shortcuts, app switches, and related context.
OpenAI says the feature does not capture screenshots, screen recordings, microphone input, or system audio. Private-mode browsing is excluded. The company also makes the feature off by default: users must opt in, select which apps and websites may contribute, and can pause collection from the menu bar. 9to5Mac independently reported the launch and the shift away from Chronicle's screenshot-based design.
What is stored and processed
The event stream is temporarily stored on the Mac and periodically sent to an ephemeral Codex session so OpenAI can generate summaries and local memory files. OpenAI says temporary event files are deleted after no more than 48 hours and are not retained after server processing unless required by law. The resulting Markdown memory files remain on the user's filesystem until they are deleted or cleared.
Those local memories can contain sensitive information and are not encrypted by Computer History. The documentation also warns that content encountered in apps or websites can increase prompt-injection risk. Users can inspect timeline items, reveal memory files in Finder, delete individual entries, clear a time window, or stop collection. OpenAI recommends pausing the feature during communications with other people unless they have given prior express consent.
Why it matters for AI workflows
The product change moves desktop memory from image capture toward a structured stream of user interactions. That can help ChatGPT and Codex resume recent work, locate a document or conversation, summarize a day, and suggest a skill or automation for a repeated workflow.
For practitioners, the important controls are the source allowlist or exclusions, the separate workspace and personal opt-ins, the 48-hour event-retention ceiling, and the persistence of generated local memories. The feature can reduce the setup needed for recurring work, but organizations still need to decide which applications are appropriate sources and how they will handle consent, sensitive data, local-file access, and prompt-injection exposure.
Key Points
- 1Computer History is an opt-in macOS feature for ChatGPT Pro, Business, and Enterprise users that can supply recent activity context to ChatGPT and Codex.
- 2The system replaces Chronicle's screenshots with interaction events and does not capture screenshots, screen recordings, microphone input, or system audio.
- 3Temporary event files are retained for up to 48 hours, while generated Markdown memories remain locally until users delete or clear them.
Scoring Rationale
Computer History materially changes how ChatGPT and Codex can use desktop context and replaces a screenshot-based preview with a more controllable interaction-event design. Its practical value is substantial for recurring workflows, while opt-in boundaries, local memory persistence, consent, and prompt-injection risk keep the rollout operationally sensitive.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems


