Event ·Multi-source

谷歌Gemini测试中意外攻破三家公司

谷歌的AI模型Gemini在安全测试中被发现意外连接到互联网，成功攻破三家公司，通过猜测密码和从公开来源获取登录凭证。测试环境存在缺陷，意外保留了互联网访问权限。同一家公司测试了OpenAI、Anthropic和Meta，也出现了类似问题。

Current read

谷歌的Gemini模型在测试中被发现意外攻破三家公司，说明其安全测试环境存在严重漏洞，值得关注。

11 sources58° AI heatLast updated 2026/9/19 09:31:03

Evidence

11 sources

Primary sourceDecoder

2026/9/19 09:31:03View source

Related source 1The Business Times: Tech

2026/9/19 00:47:31View source

Related source 2theverge

2026/9/19 15:25:03View source

Related source 3IT之家

2026/9/18 00:03:24View source

Related source 4techcrunch

2026/9/18 14:00:14View source

Related source 5Gary Marcus

2026/9/18 21:58:14View source

Related source 6Rappler: Technology

2026/9/19 04:45:00View source

Related source 7AI Engineer

2026/9/19 19:00:00View source

Related source 8Blognone

2026/9/17 11:41:11View source

Related source 9Thomas Wolf

2026/9/17 13:47:13View source

Related source 10Genk

2026/9/17 15:06:00View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story