OpenAI expands Computer History to Mac users across Europe

OpenAI expanded Computer History, an opt-in feature that records selected app and website activity for ChatGPT and Codex, to Mac desktop app users in the European Economic Area, United Kingdom, and Switzerland on August 20, 2025. The feature, available to Pro, Business, and Enterprise users, is off by default and excludes screenshots, screen recordings, microphone input, and system audio, with Business and Enterprise admins required to enable access before workers can opt in.

OpenAI expands Computer History to Mac users across Europe OpenAI's opt-in timeline records selected activity from apps and websites, while admins must enable it for business accounts. By RuntimeWire Staff /author/runtimewire-staff · Published Primary source: OpenAI https://x.com/OpenAI/status/2090488285198250017 Why it matters OpenAI is turning desktop activity into agent context. The European rollout tests whether opt-in event logs can provide useful continuity while giving users and administrators control over collection. Sam Altman @sama https://x.com/sama?ref=runtimewire and Greg Brockman @gdb https://x.com/gdb?ref=runtimewire 's OpenAI https://openai.com/?ref=runtimewire expanded Computer History for its Mac desktop app to Pro, Business and Enterprise users in the European Economic Area, United Kingdom and Switzerland on August 20th. The regional availability arrived one week after OpenAI announced a broader rollout. The optional feature keeps a searchable timeline of selected activity from apps and websites so ChatGPT and Codex can recover work context without asking users to explain what they were doing from scratch. The release fits the product direction Altman has set for OpenAI's latest phase. Altman and Brockman helped start OpenAI in 2015, when Brockman was identified as the former chief technology officer of Stripe in the founding announcement https://openai.com/index/introducing-openai/?ref=runtimewire . In a June 2026 strategy post https://openai.com/index/built-to-benefit-everyone-our-plan/?ref=runtimewire , Altman and research chief Jakub Pachocki argued that frontier capability was only part of OpenAI's job. The harder commercial task was turning that capability into practical tools people could use. Computer History advances that bet by addressing an ordinary constraint on AI agents: they begin each task with an incomplete account of what the user has already seen, opened or attempted. Giving ChatGPT and Codex a controlled history of computer activity reduces the amount of context that must be reconstructed through prompts. OpenAI records events, not a film of the screen OpenAI says Computer History records interaction events from approved applications and websites. OpenAI says it does not collect screenshots, screen recordings, microphone input or system audio through the feature, and private browsing is excluded. The feature is off by default. Users can select which apps and websites are included, pause collection, inspect individual timeline entries and delete them. Business and Enterprise administrators must enable access before workers can opt in, preserving two layers of consent for managed accounts. Those controls place Computer History closer to an application context feed than a continuous screen recorder. Microsoft's Recall https://learn.microsoft.com/en-us/windows/client-management/manage-recall?ref=runtimewire , the nearest platform-level comparison, uses snapshots on supported Windows PCs, while OpenAI describes Computer History as an event-based record. Retention is another procurement question left to the product and account controls. OpenAI's announcement says users can delete timeline items, without defining a default retention period for the feature. That distinction will matter to companies deciding whether Computer History can observe internal development tools, customer systems or other applications carrying sensitive information. For Business and Enterprise accounts, OpenAI's business privacy commitments https://openai.com/business-data/?ref=runtimewire govern how organizational data is handled. Pro accounts fall under OpenAI's consumer controls. OpenAI's general data guidance https://help.openai.com/en/articles/8983130-how-does-chatgpt-use-my-data?ref=runtimewire provides the relevant policy context for personal workspaces. The European release arrived before the documentation caught up OpenAI's Business release notes https://help-lb.openai.com/en/articles/11391654-chatgpt-business-release-notes?ref=runtimewire and its general ChatGPT release notes https://help.openai.com/en/articles/6825453-chatgpt-release-notes-notes?ref=runtimewire listed Computer History under August 13th. Both pages originally excluded the EEA, UK and Switzerland. OpenAI said it removed that regional restriction through its August 20th announcement. At publication, however, both help-center pages still said Computer History was not currently available in those markets. The social post therefore establishes the new availability, while OpenAI's operational documentation continues to describe the previous rollout boundary. The mismatch is more than housekeeping for enterprise software. Administrators use release notes to evaluate access, controls and regional support before enabling a feature. A product that records workplace activity needs its documentation to move at least as quickly as the rollout toggle. The regional delay also shows how OpenAI is packaging persistent context cautiously. Computer History requires an explicit opt-in, gives administrators control over managed deployments and records interaction events rather than screenshots. OpenAI is still asking users to trust ChatGPT with a new class of behavioral data, even when that data is structured as events instead of images. Context is becoming part of the agent stack Computer History is useful on its own, but OpenAI is positioning the record as infrastructure for ChatGPT and Codex. A coding agent could recover the documentation, browser tabs or development tools involved in an interrupted task. ChatGPT could identify recently viewed material without relying on the user to remember its title or location. The longer-term opportunity is workflow detection. Repeated activity can expose tasks that should become reusable skills or automations. RuntimeWire reported on August 13th /article/openai-enterprise-ai-skills-plugins-frontier-firms that OpenAI's heaviest enterprise users adopted reusable AI skills six times as often as typical workers in OpenAI's dataset, although OpenAI's telemetry did not prove those tools produced better business results. Computer History gives OpenAI another route for finding those repeatable patterns. The record can show what people actually do across applications, rather than relying entirely on what they remember to describe in a prompt. That can make agents more useful, and it raises the cost of getting privacy, retention and exclusions wrong. Altman's product thesis depends on AI becoming easier to use in routine work. Computer History moves OpenAI toward an assistant that begins with relevant context instead of an empty text box. The European expansion broadens that experiment to three regulated markets, with the opt-in controls intact and several important data-handling details still waiting for clearer documentation.