OpenAI has expanded ChatGPT with a new optional desktop feature called Computer History, which aims to make the AI assistant more useful by recording a user's interactions with apps and websites. The feature is rolling out to Mac users subscribed to ChatGPT Pro, Business, or Enterprise, and it is designed to create persistent memories and a searchable timeline from operating-system activity and accessibility data.

• Feature name: Computer History
• Availability: Desktop Mac app for ChatGPT Pro, Business, and Enterprise
• Function: Records interaction events across selected websites and apps to create memories and a timeline
• Privacy stance: Opt-in, no screen capture, no audio recording, with per-app and per-website controls
• Excluded regions: United Kingdom, Switzerland, and the European Economic Area
• Successor: Replaces the earlier Chronicle research preview

What is Computer History?

Computer History works in the background of the ChatGPT desktop app. It logs interaction events such as opening documents, visiting websites, or switching between applications, and it converts those events into structured memories. Those memories are then used to answer questions, provide context, and generate personalized suggestions. A user might ask ChatGPT what they were working on yesterday, which websites they visited earlier, or which files they had open during a specific project, and the assistant can use the recorded activity data to provide an answer.

Unlike traditional memory features that rely on manually supplied facts or preferences, Computer History is observational. It learns by watching what the user actually does on their computer. Over time, it can identify recurring workflows, common tools, and frequent tasks across different apps. OpenAI says that pattern recognition could lead to recommendations for skills and automated workflows that streamline a person's routine.

Rollout and availability

Initially, Computer History is limited to the ChatGPT desktop app for macOS. It is available to subscribers on Pro, Business, and Enterprise plans, and there is no mention of a free-tier rollout or a Windows desktop version at launch. The feature also will not be available in the United Kingdom, Switzerland, or the European Economic Area. OpenAI says it is replacing the earlier Chronicle research preview, suggesting that the company has been working toward this kind of persistent context technology for some time.

This announcement follows recent upgrades to OpenAI's GPT-5.6 models, which improved accuracy and reasoning adaptability while reducing unnecessary information. The new feature is part of a broader push to make ChatGPT more agentic and context-aware.

How it compares to Windows Recall

Longtime observers may see similarities between Computer History and Microsoft's Windows Recall. Recall was introduced as a Windows AI tool that captured screenshots at regular intervals so that users could later search their screen history. It caused widespread concern when it was announced, partly because it was initially switched on by default and because screenshots could include passwords, private messages, and other sensitive items. Microsoft delayed the rollout and made changes to address security issues before reintroducing it.

Computer History differs in several significant ways. It is opt-in, not on by default. It does not take screenshots or record audio. According to OpenAI, it relies on operating-system activity and accessibility data, which is a less invasive method of logging what a person is doing. Users can also control which apps and websites are allowed to contribute interaction events. These choices make the feature less alarming than Recall, at least on paper.

However, the feature still carries privacy risks. OpenAI's own announcement includes a disclaimer that Computer History can contain sensitive information. The data captured by the feature is not encrypted specifically by Computer History, and other programs running on the same computer could potentially access it. That means a malicious app or unauthorized process might be able to read the recorded activity if it can reach the data store. This is a concern for anyone who works with confidential documents, personal communications, or financial information.

Privacy and security implications

For security-conscious users, the opt-in nature is helpful but not a complete shield. The data lives on a device in an area that may be accessible to other software. OpenAI does not say whether the data is encrypted at rest inside the local data store or whether it is transmitted to servers. The detailed timeline could also become a target for anyone with remote access to the machine. Users should consider whether they are comfortable with a record of every app and website visit being stored in a format that another process could read.

Data retention and control also matter. If the feature is turned off, does it immediately delete existing memories? Can users edit or individually remove entries from the timeline? OpenAI has not provided a full set of controls in the initial announcement. The company notes that users can choose which apps and websites contribute to interaction events, but it does not say whether users can pause the collection process for short periods or erase specific memories. These questions are likely to emerge as more people begin testing the feature.

The broader shift toward AI memory

The introduction of Computer History reflects a growing trend in AI. Assistants are moving from simple Q&A chatbots to proactive agents that remember context, anticipate needs, and perform actions on behalf of users. ChatGPT already has memory functions that let users save facts and preferences. Google has experimented with personalization across its services, and Apple has been developing on-device intelligence that can summarize and retrieve information without sending everything to the cloud.

The key differentiator with Computer History is that it is built around fine-grained activity logging across separate apps and websites. This can create a rich model of a person's digital life, but it also raises the stakes for privacy breaches. A chatbot with a detailed record of a user's browsing habits, document edits, and communication patterns could be extraordinarily useful for personalization, but it could equally become a source of deep surveillance if the data falls into the wrong hands.

Regulatory and regional concerns

The exclusion of the United Kingdom, Switzerland, and the European Economic Area is notable. European data protection laws, including the General Data Protection Regulation, place strict limits on large-scale collection and processing of personal data. Features that continuously log user activity require strong justifications, clear consent, and robust data protection safeguards. EEA countries also have special protections for employees who use company-issued devices, which might make a feature like Computer History difficult to introduce without violating workplace privacy expectations.

OpenAI's decision to exclude those regions suggests the company may be positioning the feature to avoid immediate regulatory scrutiny or is still working on a compliance-friendly version. Whether the feature eventually reaches those markets will depend on how the company addresses consent, retention, and the right to erasure.

What users should consider before enabling

Before enabling Computer History, users should think about what kind of information they regularly handle. Anyone who deals with trade secrets, medical records, legal advice, or personal conversations with clients should be cautious. Even if the feature is opt-in, its background logging might create records that the user does not intend to keep. Users should also talk with their employers before enabling it on work computers, especially if they are on a Business or Enterprise plan and the organization has its own data handling policies.

From a productivity standpoint, Computer History could be convenient. It could help people who work across many projects and applications because they would no longer need to remember exactly where each task stopped. Asking ChatGPT to summarize a previous session or locate a particular file could save time. The feature might also help with onboarding, research, creative brainstorming, and project handoffs.

But the convenience comes with trade-offs. With a continuous activity record, there is no way to keep some activities completely invisible from the assistant unless users carefully set exclusions. It is also unclear how long the memories persist and whether they can be exported or permanently purged. These details matter for high-stakes environments.

What comes next

OpenAI has not announced when Computer History might come to Windows or other platforms. The Mac launch is likely a controlled rollout to test user reactions and gather feedback. If the feature gains traction, OpenAI could expand it to additional plans, regions, and devices. But if privacy concerns dominate the conversation, the company may be forced to add more safeguards or place tighter limits on what gets recorded.

The coming months will likely reveal whether users are ready to hand ChatGPT a front-row seat to their daily computing lives. The feature is a significant step toward truly context-aware AI, but it also sits at the center of a long-running debate about the line between useful memory and persistent surveillance.

Source: Android Authority News