OpenAI has introduced Computer History, a new feature of the ChatGPT desktop app for macOS. It records clicks, keystrokes and on-screen activity, building a timeline that ChatGPT and Codex can consult. The stated goal is to suggest automations and pick up tasks left half-finished.
Still, the implications for company privacy are far from trivial. The feature is opt-in, which is a good thing. But in professional settings, the line between personal productivity and sensitive company data is thin. That's why IT managers and marketing managers need to think carefully before enabling it on company devices. You can exclude individual apps and websites, and delete single entries from the history.
In short, we at SHM Studio see Computer History as a significant step toward agentic AI built into the daily workflow. Before adopting it in business environments, though, you need a clear policy on which applications can be monitored. The opportunities for marketing automation are real, but they call for governance.
What changed with ChatGPT's Computer History
On August 16, 2026, OpenAI released Computer History for the ChatGPT desktop app on macOS. The feature records user activity: clicks, keystrokes, open windows and content viewed. From this data, it builds a chronological timeline the model can consult. According to The Verge, both ChatGPT and Codex can access this history to answer context-based requests.
Ari Weinstein, Product and Engineering Manager at OpenAI, clarified on X that the feature automatically ignores tabs in private or incognito mode. Users can also exclude specific apps and websites from monitoring, and delete individual entries from the history. So granular control exists, but it requires active setup.
Making the feature opt-in matters. Unlike many telemetry features that run by default, here users have to turn it on knowingly. This approach lowers the risk of accidental exposure, but doesn't eliminate it entirely in a business context.
How the timeline and the suggestion engine work
Computer History doesn't just record passively. The system analyzes user behavior patterns to identify recurring tasks. It then proposes automations based on sequences of actions already performed. For example, if a user regularly copies data from an Excel sheet into a CRM, the system can suggest a Codex script to automate the flow.
The feature is also designed to resume unfinished tasks. If a user started writing a report and stopped midway, ChatGPT can offer to pick up where they left off. This mechanism comes close to the concept of agentic AI: a model that doesn't just respond to explicit prompts, but anticipates what the user needs.
To dive deeper into AI agent architecture and its operational implications, we at SHM Studio recommend exploring how these developments fit into a broader strategic picture.
Immediate impact on company privacy
For marketing and IT leaders at SMEs and mid-market companies, this feature raises concrete questions. The first concerns data sovereignty: what happens to the information the timeline records? OpenAI has not yet published detailed technical documentation on retention policies or on whether this data is used to train its models.
So before enabling Computer History on company devices, a few points are worth checking. First, if your company handles personal data of customers or employees, monitoring desktop activity could amount to additional data processing under the GDPR. In particular, companies based in the EU need to assess whether this use falls within the legal bases already declared in their record of processing activities.
You should also consider the risk of exposing confidential information — contracts, financial data, internal communications — to the timeline of a cloud service. Being able to exclude individual apps is useful, but it requires mapping sensitive applications first. At SHM Studio, we suggest drafting an internal policy before any rollout.
The open front: governance and company policies
The real challenge isn't technical. It's organizational. Many Italian companies still lack a structured policy on using AI tools at work. Computer History makes that gap more urgent. A feature that records all desktop activity expands the risk surface exponentially compared with a simple text chatbot.
So what concrete steps should you take? First, take stock of the company devices where the ChatGPT app for macOS is installed. Then decide whether to enable, disable or selectively configure Computer History by role or department. Finally, update your acceptable use policies to explicitly cover this type of tool.
If you manage LinkedIn campaigns or Google Ads campaigns, the issue is especially delicate: advertising platforms handle audience data that must not be exposed to unauthorized monitoring tools.
Opportunities for marketing automation: a practical look
Risks aside, Computer History opens up interesting scenarios for anyone working in hands-on marketing. The system's ability to spot repetitive patterns and suggest automations can cut the time spent on low-value manual tasks. Think of generating periodic reports, updating dashboards or filling in creative briefs.
In particular, SEO and copywriting teams could benefit from an assistant that remembers the context of ongoing projects. Instead of rewriting the brief for ChatGPT every time, the model could already have the context it needs thanks to the timeline. This reduces cognitive load and speeds up content production cycles.
Still, it's important not to overestimate current capabilities: generative models remain support tools, not a substitute for human judgment. The quality of the suggested automations will depend on the quality and consistency of the recorded behavior.
Outlook: toward contextual AI built into the workflow
Computer History points the way, but it isn't a finished product. OpenAI is building an AI model that doesn't live only in the browser or the chat app, but is woven into the operating system and the daily workflow. This direction is consistent with what OpenAI has announced in recent months about the agentic capabilities of GPT-4o and its successors.
For Italian companies, the strategic question is: when this technology is mature and safe for enterprise use, will we be ready to integrate it? The answer depends on the quality of your existing digital infrastructure. Companies that have already invested in solid web architectures and structured digital marketing processes will have a competitive edge in adoption.
So the right time to prepare isn't when the technology is already widespread. It's now, while the market is still defining its standards. If you want to explore the strategic implications of AI for your business, you can contact SHM Studio for an initial assessment. Our blog also collects up-to-date analysis on these topics, to help you find your way in a fast-changing landscape.
In short, Computer History is a feature to watch closely. The opportunities are there, but they call for a governance framework before any adoption in professional environments. SHM Studio's AI consulting is available to support you along the way.
Related articles
Discover more articles exploring similar topics, selected to offer you a more complete and stimulating perspective. Each piece of content is carefully chosen to enrich your experience.
