HomeBusinessChatGPT Computer History Raises New Privacy Questions for Work Macs

ChatGPT Computer History Raises New Privacy Questions for Work Macs

OpenAI has introduced a feature called Computer History to the ChatGPT Mac app, according to ZDNET’s reporting on the rollout. The opt-in tool builds a timeline from activity across the apps and websites a person allows on a Mac, with access initially reported for Pro, Business and Enterprise ChatGPT accounts in most regions.

For businesses, the issue is not only whether the feature can help users resume unfinished work, locate documents or summarize tasks. It is what happens when a timeline is built from activity on a device that touches customer records, contracts, financial systems or proprietary code. The feature is described as user-controlled, but the records it creates may still include sensitive business context if employees enable it on work Macs.

The available reporting describes what Computer History captures and the controls OpenAI provides. It does not establish how the feature performs at scale inside a business, or how organizations should weigh productivity against privacy for their own workflows. Businesses should therefore treat the rollout as a governance question, not simply a productivity feature.

Computer History changes what the Mac app can remember

Computer History replaces an earlier ChatGPT feature called Chronicle, which monitored screen activity through screenshots, according to ZDNET. The newer approach reportedly records interaction events rather than taking screenshots, and OpenAI says it does not capture the screen, microphone input or system audio.

Instead, the feature uses macOS accessibility capabilities to observe activity from apps and websites the user includes. The resulting event stream can reportedly include clicks, typing, keyboard shortcuts, app switches and context exposed through the accessibility system. Those events are periodically converted into text summaries and local memory files that populate a timeline organized by day and time.

Each entry can include a title, an activity summary, the apps or websites involved, and a suggested skill for repeatable work. Users can view the associated memory file in Finder or delete individual entries. To enable the feature, users open Settings, select Integrations and turn on Computer History, with Memories also required for activity to be used across chats and tasks.

Account type matters. A Pro subscriber can reportedly enable the feature independently, while Business and Enterprise accounts require an administrator to grant access before employees can opt in. Users can select allowed apps and websites, pause collection, resume it, turn the feature off, clear recorded history and delete individual entries.

Work Macs can expose more than a personal productivity trail

The practical concern is what a work Mac may expose once Computer History is enabled. A timeline built from activity in email, a browser tab showing a client record, or a code editor used for unreleased software can turn routine work into a structured, searchable record.

A person who uses the same Mac for tax software, client communications and personal browsing could create a timeline that brings business and personal activity together. Similar issues may arise when allowed apps or websites include legal-document platforms, healthcare scheduling systems, customer-relationship-management tools, internal dashboards or financial applications.

Business and Enterprise accounts include an administrator approval step before employees can opt in, which gives organizations a chance to decide whether the feature should be available and which workplace systems should be excluded. That control is only useful if administrators and employees understand the sensitivity of the apps and websites being added.

Local memory files create a specific governance issue

The reported documentation says Computer History files can contain sensitive information, are not encrypted by Computer History itself and may be accessible to other programs running under the same macOS user account. That disclosure does not establish that a breach has occurred or that the files are more exposed than other unencrypted local data, but it does identify a concrete risk businesses should assess.

Supplementary reporting says raw interaction-event files remain on the Mac in the ChatGPT app’s data container and are deleted after 48 hours, while OpenAI processes the event files on its servers to produce summaries, does not retain those raw files afterward and does not use them for training. The plain-text memory files remain until the user deletes them.

OpenAI’s official macOS app help pages confirm that ChatGPT desktop app data generally follows the same account-level data policies as the web version, while macOS-specific behavior can vary by feature. Because Computer History is a new, opt-in capability described through product reporting, businesses should review current OpenAI documentation before enabling it across managed devices.

Businesses should review controls before enabling the feature

Small businesses considering Computer History, or discovering that employees have already enabled it, can take several concrete steps before deciding how broadly to allow it.

  • Identify Macs running ChatGPT and determine whether each device is connected to a personal, Pro, Business or Enterprise account.
  • Use administrator approval on managed Business and Enterprise accounts before employees can opt in.
  • Decide whether personal and business activity should be separated across devices or accounts.
  • Test the feature only with low-sensitivity apps and websites before adding systems that display customer records, financial information or proprietary code.
  • Protect the relevant macOS user account, since reported memory files may be readable by other programs running under the same account.
  • Explain pause, deletion and source-selection controls to employees before rollout.
  • Document which devices and workflows are approved for use with Computer History and why.

Productivity benefits should be measured against privacy exposure

The feature could be useful for employees who frequently return to unfinished work, need a summary of recent activity or want repeatable skills built from their own workflows. Those are legitimate productivity uses, especially for solo operators and small teams without dedicated process documentation.

The same feature can also create a new record of activity that a business did not previously maintain. For a company handling health records, financial data, legal work, customer contracts or proprietary software, the relevant question is not whether the timeline is convenient. It is whether the organization is comfortable with the sources included, where summaries are stored, how long they remain and who could access them.

That makes Computer History a data-governance issue as much as a user-experience feature. Businesses should avoid treating opt-in status as the only safeguard, because an employee may not be in the best position to judge whether an app contains sensitive information that should be excluded.

Further information about the feature’s storage and security model could change how businesses assess Computer History. Independent testing of local storage behavior, account-access protections and deletion controls would provide evidence beyond product descriptions and early reporting.

The reported rollout to the United Kingdom, Switzerland and the European Economic Area is also worth tracking because those regions were not included in the initial availability described by ZDNET. Changes to documentation, retention details or business-admin controls should be reviewed before any wider workplace deployment.

Convenience does not remove the need for data governance

Computer History appears more limited than the screenshot-based Chronicle feature it replaces, and the reported controls include opt-in activation, administrator approval for managed accounts, selective app inclusion, pausing, clearing history and deletion. Those controls do not support treating the feature as categorically unsafe.

They also do not remove the governance questions raised by AI tools that interact with workplace devices: what the feature can observe, which sources are allowed, where records are stored and who might be able to access them. The practical takeaway is measured. Businesses can test the feature deliberately, exclude sensitive workflows and protect relevant Mac accounts rather than responding automatically to a new opt-in prompt.

 

Must Read

spot_img