Gemini Spark 確實有效 —— 為何隱私風險如此之高？
the verge 測試了谷歌的 'gemini spark' 智慧體，發現它能成功執行起草郵件和分析私人預算表等複雜多步驟任務。這種動手能力證實了自主智慧體的興起，但也引入了關於個人數據訪問的重大未解決隱私問題。
the verge tested google's new 'gemini spark' agent, revealing it successfully completed complex, multi-step tasks such as identifying a spouse from contacts, extracting data from unlabeled budget spreadsheets, and drafting personalized emails. the agent operates autonomously in the background, demonstrated by its ability to retroactively insert links into previously drafted text. despite some failures in complex planning scenarios, the system represents a significant step toward practical autonomous agents. however, the review highlights that this capability requires deep access to personal data, raising substantial privacy and security questions for users.