TLDRocket
Sign in

OpenAI added an optional macOS “Computer History” feature to ChatGPT Work and Codex that records user interactions to build a timeline of recent app and website activity

Feature update Updated 68% confidence first seen

OpenAI released an optional “Computer History” feature for ChatGPT Work and Codex on macOS. The feature creates a timeline of user interactions (including clicks and keystrokes) that ChatGPT can use to answer questions about recent work, with controls to exclude certain apps or websites and to delete stored entries. Coverage raises concerns about how keystroke data is handled, including claims it is not encrypted.

Decision brief

What changed
OpenAI added an optional macOS feature called Computer History to ChatGPT Work and Codex that records recent app and website interaction events, including clicks and keystrokes, to build a timeline ChatGPT can use to answer questions about recent work. The feature is off by default, with controls to exclude specific apps or sites and to review or delete stored entries.
Why it matters
This changes the tradeoff between AI convenience and endpoint data exposure: teams can reduce context-switching and repeated prompting, but only by allowing a local activity log of user behavior that may include sensitive work patterns and typed input. For business leaders, the decision is less about model capability than about whether existing privacy, endpoint management, and acceptable-use controls are strong enough to govern an opt-in tool that observes employee activity at this level. If adopted, it may also affect employee trust and internal policy requirements for monitoring-like features, especially where regulated or confidential data is handled.
Affected roles
CEO COO CTO CISO
Evidence
The New Stack and The Verge both report that OpenAI introduced an optional macOS Computer History feature for ChatGPT/ Codex that records interaction events, is off by default, and includes exclusion and deletion controls. The Neuron separately reports that keystroke data is stored unencrypted, but that specific security claim is not independently confirmed in the other two cited articles.
What remains uncertain
It remains unclear exactly which interaction data fields are stored, how long they persist, whether any data ever leaves the device, and whether enterprise admins can centrally disable or govern the feature. The reported lack of encryption comes from one outlet in this coverage set and should be treated as unverified here until OpenAI documentation or broader independent reporting confirms the implementation details.
Monitor next
Watch for OpenAI technical documentation or enterprise admin controls that specify storage architecture, encryption status, retention, and policy management for Computer History.

Analytical support, not advice — assumptions and open questions stated above.

Source coverage

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.