How does OpenAI’s Computer History feature for the ChatGPT macOS app work, including what it captures, where and for how long it stores dataIllustration, not a screenshot of the Computer History interface.
AI 提示
Create a landscape editorial hero image for this Studio Global article: How does OpenAI’s Computer History feature for the ChatGPT macOS app work, including what it captures, where and for how long it stores data. Article summary: Computer History is an opt-in feature in the ChatGPT macOS app that turns activity from selected apps and websites into a timeline and memories ChatGPT and Codex can use. Its main trade-off is that the resulting memory f. Topic tags: general, documentation, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fak
openai.com
想 ChatGPT 或 Codex 接住你喺 Mac 上做到一半嘅工作,而唔使次次重新交代背景?Computer History 就係為咗提供呢種上下文。不過,值得留意嘅唔係「佢有冇長期錄影」,而係佢會唔會將你嘅活動整理成可讀、持續保存嘅記憶檔案。21021
佢實際記錄乜?
Computer History 只從你允許嘅 App 同網站建立操作事件紀錄,包括點擊、輸入文字、鍵盤快捷鍵、切換 App,以及 macOS 輔助使用功能提供嘅上下文。OpenAI 表示,佢唔會擷取螢幕截圖、螢幕錄影、咪高峰輸入或系統聲音,私人瀏覽亦唔包括在內。但就算冇影像,操作事件仍可能透露你處理緊嘅敏感事情。210
「最多 48 小時」唔代表所有資料都會消失
要分清兩類檔案:
原始操作事件:暫存在 Mac 上 ChatGPT App 嘅本機資料容器,最多 48 小時後刪除。事件會經處理、整理成摘要;「原始檔存在本機」唔等於 OpenAI 系統完全唔會處理當中資訊。2122
Computer History 唔會喺功能層面加密記憶檔案。安全報道指出,以同一個 macOS 用戶帳戶運行嘅其他程式,可能毋須管理員權限就讀到檔案。如果帳戶已被資訊竊取型惡意程式入侵,呢啲檔案就可能變成一份持續保存嘅工作活動摘要——即使短暫保留嘅原始事件早已刪除。呢個係入侵後增加嘅資料外洩風險,唔係話開啟功能本身會安裝惡意程式。1821
另一個風險係提示注入:獲准網站或 App 嘅內容可能夾雜寫畀 AI 睇嘅惡意指令。如果指令混入記憶,再成為日後對話嘅上下文,ChatGPT 或 Codex 可能錯當成應該執行嘅要求。OpenAI 亦建議,盡量只畀 AI 代理接觸完成任務所需嘅敏感資料同憑證。172116