In the quiet hum of everyday computing, OpenAI has introduced a feature called Computer History that records the keystrokes and mouse clicks of Mac users, feeding this stream of digital behavior to ChatGPT as context for future conversations. Positioned as a more restrained alternative to Microsoft's screenshot-based Recall tool, the feature nonetheless raises enduring questions about where the boundary between assistance and surveillance truly lies. As AI systems grow closer to the intimate rhythms of our daily work, humanity is being asked — often without fully realizing it — how much of its
ChatGPT's Computer History feature logs keystrokes and clicks on Mac devices
A keystroke log is a more direct window into someone's digital life
Mark
So OpenAI is logging keystrokes now. How is that different from what Microsoft tried to do with Recall?
Mimi
The technical mechanism is different. Recall captures screenshots—visual snapshots of your screen. Computer History logs the actual keystrokes and clicks. OpenAI's argument is that this is less intrusive because it doesn't capture sensitive visual information.
Mark
But doesn't a keystroke log tell you everything? Every password, every search, every private message?
Mimi
Exactly. In some ways it's more revealing than a screenshot. A screenshot is a moment in time. A keystroke log is a continuous record of what someone is thinking and doing. The privacy calculus isn't as clean as OpenAI suggests.
Mark
Why would anyone turn this on?
Mimi
Context. If ChatGPT knows what you're working on, it can give better answers. It's genuinely useful. The trade-off is that you're giving the system—and by extension, OpenAI—a detailed record of your activity.
Mark
Is the data encrypted?
Mimi
That's one of the key unknowns right now. We don't have clear answers about how the data is stored, who can access it, or how long it's kept. Those details will determine whether this is actually safer than Recall or just a different kind of risk.
Mark
What happens next?
Mimi
Watch whether this becomes opt-in or default, and whether it spreads to other platforms. If it becomes standard, we're looking at a fundamental shift in how AI systems monitor our behavior.
El Pulso
- OpenAI's new Computer History feature silently logs every keystroke and mouse click on Mac devices, handing ChatGPT a running record of a user's digital life.
- The plain-text storage of this data creates serious vulnerabilities — passwords, private messages, and search queries are all captured in a form that could be devastating if compromised.
- OpenAI frames the tool as a privacy improvement over Microsoft's Recall, which sparked fierce backlash for continuously screenshotting user activity, but critics argue keystroke logs may be even more intimate than images.
- Key questions about data retention, logging granularity, and security safeguards remain unanswered, leaving users unable to fully assess the risks they are accepting.
- The precedent being established on Mac today is likely to expand to Windows, iOS, and Android — making the norms set now consequential for how AI inhabits our devices for years to come.
In the quiet hum of everyday computing, OpenAI has introduced a feature called Computer History that records the keystrokes and mouse clicks of Mac users, feeding this stream of digital behavior to ChatGPT as context for future conversations. Positioned as a more restrained alternative to Microsoft's screenshot-based Recall tool, the feature nonetheless raises enduring questions about where the boundary between assistance and surveillance truly lies. As AI systems grow closer to the intimate rhythms of our daily work, humanity is being asked — often without fully realizing it — how much of itself it is willing to share in exchange for convenience.
OpenAI has launched a feature called Computer History for ChatGPT that records keystrokes and mouse clicks on Mac computers, storing this activity in plain text so the AI can draw on it as context across conversations. The company presents it as a more privacy-conscious approach than Microsoft's Recall tool, which captured continuous screenshots of user activity and triggered swift, widespread backlash — enough to force Microsoft to delay the rollout and make it opt-in.
The distinction OpenAI is drawing has some technical merit. Keystroke logs avoid capturing sensitive visual elements like on-screen passwords or private images. But the privacy trade-off is not as clean as the company's framing implies. A plain-text record of everything a person types — every search, every draft message, every credential — is in many ways a more direct and exploitable window into someone's digital life than a photograph of their screen.
What remains unknown is how long OpenAI retains this data, how granular the logging actually is, and what protections exist against breaches or unauthorized access. These unanswered questions are what separate a genuinely less invasive tool from a different version of the same problem. Users who enable Computer History are effectively giving ChatGPT a continuous commentary on their behavior, which the AI uses to personalize and remember — a capability that is useful precisely because it is so revealing.
For now, the feature is limited to Mac devices, but the trajectory is clear. If it gains traction or becomes a default, similar tools will follow on every major platform. The deeper question the moment is posing is not which form of monitoring is marginally less intrusive — it is whether the quiet normalization of either should be accepted at all.
OpenAI has rolled out a new feature for ChatGPT called Computer History that records everything you type and click on your Mac. The system stores this activity data in plain text, making it available to the AI as context for future conversations. The company frames this as a privacy-conscious alternative to screenshot-based monitoring—a direct response to the backlash Microsoft faced over its Recall tool, which captured full screenshots of user activity and sparked immediate concerns about surveillance and data exposure.
The distinction OpenAI is drawing matters, at least in theory. Rather than taking pictures of your screen, Computer History logs the actual keystrokes and mouse movements themselves. The company's reasoning is straightforward: keystroke logs are less visually intrusive than screenshots, and they give ChatGPT the information it needs to understand what you're working on without capturing sensitive visual elements like passwords or private images that might appear on screen.
But the privacy calculus here is more complicated than OpenAI's framing suggests. Storing keystrokes in plain text creates its own set of vulnerabilities. Every password you type, every search query, every private message draft—all of it becomes data that the system can access. If that data is compromised, either through a breach or through unauthorized access, the consequences could be severe. A keystroke log is, in many ways, a more direct window into someone's digital life than a screenshot, which at least requires someone to actually look at the image.
The feature arrives at a moment when tech companies are racing to integrate AI into everyday computing. Microsoft's Recall tool, announced earlier this year, was designed to let users search through their entire computer history using natural language queries. The idea appealed to some users—imagine being able to ask your computer, "What was that article I read last week?" and having it instantly retrieve it. But the implementation terrified privacy advocates. The tool would capture screenshots constantly, store them locally, and make them searchable. The backlash was swift and substantial enough that Microsoft delayed the rollout and promised to make it opt-in rather than default.
OpenAI appears to have learned from Microsoft's misstep. By focusing on keystroke and click data rather than full screenshots, the company can claim it's being more thoughtful about privacy. The data is also stored in a way that ChatGPT can actually use—the AI doesn't need to see your screen; it needs to know what you're doing. This is a more elegant technical solution, even if the privacy implications remain real.
What's unclear at this stage is how granular the logging is, what safeguards exist around the data, and how long OpenAI retains it. These are the questions that will determine whether Computer History is genuinely less invasive than Recall or simply a different flavor of the same surveillance. Users enabling this feature are essentially giving ChatGPT a running commentary of their digital behavior, which the AI can then use to personalize responses and remember context across sessions. That's useful. It's also a significant expansion of what data OpenAI collects about its users.
The feature is currently available on Mac devices, which narrows its immediate reach but signals where the technology is headed. If Computer History proves popular—or if it becomes the default rather than opt-in—we can expect similar tools to appear on Windows, iOS, and Android. The precedent being set now will shape how AI systems interact with our devices for years to come. The question isn't whether keystroke logging is better or worse than screenshot surveillance. It's whether either approach should be normalized at all.
Citas Notables
OpenAI frames Computer History as a privacy-conscious alternative to screenshot-based monitoring, responding to backlash Microsoft faced over Recall— OpenAI's positioning
Quieres la nota completa?
Lee el original en Google News ↗
Technology