OpenAI has introduced a controversial new feature called Computer History that fundamentally changes how ChatGPT interacts with user devices. The feature, now available to select macOS users, automatically records extensive activity data including keystrokes, mouse clicks, keyboard shortcuts, and app switches—creating a detailed digital footprint of user behavior on their machines.
What Happened
The Computer History feature represents OpenAI’s latest attempt to enhance ChatGPT’s contextual awareness and functionality. By maintaining a searchable memory of user interactions, the AI system can theoretically provide more personalized and contextually relevant responses. However, the implementation has raised immediate red flags within the cybersecurity community.
The most concerning aspect is how OpenAI stores this sensitive data. All recorded activity is saved locally as unencrypted plain text files—meaning anyone with access to the device can read the raw keystroke logs without requiring additional authentication or decryption passwords. This storage methodology stands in stark contrast to industry-standard encryption practices that major tech companies typically employ for sensitive user data.
Notably, OpenAI has deliberately restricted Computer History availability in the European Union, Switzerland, and the United Kingdom. This geographic limitation suggests the company anticipated regulatory pushback under stricter European privacy frameworks like GDPR.
Key Points
Security experts have highlighted several critical vulnerabilities in this approach. First, unencrypted plain text storage creates a treasure trove for malicious actors who gain device access. Second, the feature captures everything typed—potentially including passwords, authentication codes, sensitive business information, and personal communications. Third, there’s no transparent mechanism showing users exactly what data is being recorded or how long it persists.
The feature’s opt-in nature provides minimal comfort, as many users may not fully understand what data collection entails before enabling it. Additionally, the absence of European availability suggests OpenAI recognizes this feature conflicts with stringent privacy regulations.
What This Means
For American users, Computer History represents a significant privacy trade-off in exchange for enhanced AI functionality. Those considering enabling this feature should carefully weigh whether improved ChatGPT responsiveness justifies exposing keystroke-level activity data on their machines.
This development also underscores the broader tension in AI development between utility and privacy. As AI systems become more sophisticated, they increasingly demand access to personal behavioral data. Companies like OpenAI face pressure to balance competitive product features against legitimate user security concerns.
Users should strongly consider disabling Computer History or adopting additional device security measures if they choose to use it. This incident serves as a reminder that convenience and advanced AI capabilities often come with hidden privacy costs that users must actively manage themselves.
