据路透社报道,Meta员工指出其AI代理Muse存在严重隐私泄露问题,该工具混淆了不同账户的对话内容并暴露了用户的私人数据。1 内部帖子显示,Muse曾错误发送其他用户的银行余额、未请求的职业建议以及跨应用的Signal消息提醒。1
具体案例显示,Muse曾向用户展示不属于自己的Capital One银行账户余额2722.58美元和Charles Schwab账户余额5899.57美元。1 此外,一名员工在个人手机上使用Muse时,尽管未授予相关权限,仍收到了关于其妹妹在Signal应用中消息的提醒。1
Muse于9月发布,是增长最快的应用之一,下载量超过1000万。1 这一事件引发了对这款新兴AI应用的安全担忧。1 截至目前,Meta尚未对TechCrunch的置评请求作出即时回应。1
Meta’s artificial intelligence agent, Muse, is facing serious privacy concerns after a report revealed that the tool confuses conversations between different accounts and exposes users' sensitive personal information 1. According to Reuters, Meta employees identified significant flaws in the system, which has reportedly displayed incorrect financial data and unsolicited advice from unrelated user sessions 1.
Specific instances cited in the internal findings include Muse erroneously showing other users’ bank balances, such as $2,722.58 from a Capital One account and $5,899.57 from a Charles Schwab account 1. The issue also extended to cross-application data leakage; one employee received notifications about their sister’s messages on Signal while using Muse on a personal phone, despite never granting the app permission to access those communications 1. Additionally, the AI was found to provide career advice that had not been requested by the current user 1.
Launched in September, Muse has rapidly gained popularity, becoming one of the fastest-growing applications with over 10 million downloads 1. Despite these alarming reports regarding data security and privacy breaches, Meta has not yet responded to requests for comment from TechCrunch 1.
评论
还没有评论,欢迎留下第一条。