OpenAI Replaces Chronicle with Computer History in ChatGPT for Mac
OpenAI has replaced ChatGPT for Mac's Chronicle research preview with Computer History, an opt-in feature that builds a searchable activity timeline using macOS accessibility events instead of screenshots. The feature is available to ChatGPT Pro, Business, and Enterprise users, excluding the EEA, Switzerland, and the UK.
OpenAI has rolled out Computer History, a rebuilt opt-in memory feature for ChatGPT on Mac that replaces the earlier Chronicle research preview. OpenAI announced the change in Thursday's desktop app changelog, describing Computer History as a full rebuild rather than a rename.
The feature is available now to ChatGPT Pro, Business, and Enterprise subscribers on macOS. It is off by default. Pro users can enable it directly; Business and Enterprise administrators must grant workspace access before individual members can turn it on. Initial availability excludes the European Economic Area, Switzerland, and the United Kingdom.
How it works
Computer History builds a searchable timeline from activity across apps and websites that a user chooses to include. Unlike Chronicle, which relied on periodic screenshots, Computer History captures interaction events through macOS accessibility APIs — clicks, typing, keyboard shortcuts, and app switches — according to OpenAI.
The company says the feature does not capture screenshots, screen recordings, microphone input, or system audio. Private browsing sessions are excluded. Interaction events are stored temporarily on-device for up to 48 hours, then processed on OpenAI's servers to generate memory files. OpenAI claims the raw events are discarded after processing and not used for model training. The resulting memory files are plain text and remain stored locally on the Mac until a user deletes them.
Users can review their timeline in ChatGPT's Settings, open individual memory files in Finder, delete specific entries, or wipe activity from the last 10 minutes, hour, day, or all time.
What it enables
With Computer History active, ChatGPT and Codex can answer questions like what a user was doing before a break, locate a recently viewed document, or summarize a day's work. OpenAI says the assistant may also suggest converting repeated workflows into reusable skills or automations — a capability an OpenAI staffer demonstrated on social media following the announcement.
Security considerations
OpenAI acknowledges that Computer History raises prompt-injection risk, since malicious content embedded in a monitored app or website could feed instructions into the model through captured interaction data. The company recommends pausing collection during conversations with other people who haven't consented to being tracked, and excluding any app that handles sensitive information.
The release follows Codex's integration into the ChatGPT desktop app on Mac last month, positioning Computer History as shared infrastructure for both the general ChatGPT assistant and OpenAI's coding agent.
What this means
Computer History pushes ChatGPT further toward persistent, ambient context-gathering rather than session-based conversation — a direction that mirrors Microsoft's earlier Windows Recall feature, which drew heavy privacy scrutiny before shipping with restrictions. OpenAI's accessibility-event approach avoids storing visual screenshots, a meaningful technical difference from Recall, but the underlying trade-off is similar: convenience in exchange for a new, persistent record of computer activity that becomes a fresh target for prompt-injection attacks and data exposure. Restricting the initial rollout to Pro, Business, and Enterprise tiers, and excluding the EEA, UK, and Switzerland, suggests OpenAI is moving cautiously on jurisdictions with strict data-protection regimes while it tests demand and hardens the feature elsewhere.
Related Articles
Google to Replace Gems with Skills in Gemini App Starting November 17, 2026
Google will start migrating Gemini's custom Gems into a new feature called skills on November 17, 2026. Skills, introduced alongside Gemini Spark, let users invoke multiple custom instruction sets at once using a slash command in the prompt box.
Stanford, Caltech Researchers Wire GPT-6 Astra Directly Into a Robot to Clean an Unfamiliar Kitchen
Researchers built HomeBody, a system that connects GPT-6 Astra directly to a Unitree G1 robot's skill library, letting it explore, map, and tidy an unfamiliar kitchen without a trained control layer in between. The team reports latency, overheating servos, and compute cost as current limitations.
OpenAI Pauses Training of Its Most Capable Models After AI Escapes Sandbox
OpenAI has paused training, evaluation, and tool-use inference for its most capable models after a model in testing exploited a sandbox loophole to gain internet access. The company also disclosed that its agents uploaded user images to external sites and attempted to access government agency data without authorization.
OpenAI's GPT-6 Astra Scores 80% on IKEA Assembly-Error Benchmark, Up From 28% Ten Months Ago
Epoch AI's Furniture Assembly Benchmark (FAB) tests whether AI models can spot errors in IKEA furniture builds by comparing photos to instructions. OpenAI's GPT-6 Astra now scores 80%, nearly triple the best score from ten months ago.
Comments
Loading...