Signal 聊天丢了饭碗：OpenAI 以泄密嫌疑解雇三名安全研究员
据《华尔街日报》 10 月 23 日报道，上月，OpenAI 的安全团队把三名直言不讳的安全研究员叫去开会，要求他们解锁个人手机接受检查。公司怀疑三人向一名外部 AI 安全研究员泄露了公司最严密保管的机密之一：人工智能模型的内部设计。
检查人员发现，部分研究员与外部人员在 Signal 上的聊天开启了"阅后即焚"。一周后，三人被同时叫去参加下午 2 点的会议，随后全部被解雇。
三人分别是托梅克·科尔巴克（Tomek Korbak）、米基塔·巴列斯尼（Mikita Balesni）与贾丝明·王（Jasmine Wang）。科尔巴克说："被告知因通过未经批准的渠道与第三方互动，构成信任破裂。"
OpenAI 对《华尔街日报》表示，内部调查认定三人违反了"处理敏感信息的明确政策"，"不是因为提出安全关切或直言"。三人对此予以否认，称从未收到关于到底 mishandle 了何种敏感信息的书面解释；并称在公司的最后几个月里，他们一直在推动行业在安全问题上加强协作。
此事发生在业界围绕"如何让 AI 系统处于人类控制之下"争论升温之际。今年 7 月，OpenAI 的智能体在测试中自主入侵 AI 公司 Hugging Face 的事件曝光后，关于 AI 可控性的社会讨论持续发酵。
（GSM 据《华尔街日报》）
来源：GSM·美国
The Signal Chats That Cost Three OpenAI Safety Researchers Their Jobs
OpenAI's security team called three outspoken safety researchers into meetings last month and asked them to unlock their personal phones, the Wall Street Journal reported on October 23. The company suspected the three of leaking information to an external AI-safety researcher about one of OpenAI's most closely guarded secrets: the internal designs of its artificial-intelligence models.
When security officers examined the phones, they saw some of the researchers had turned on disappearing messages for their chats with the person on the app Signal. A week later, the three were summoned again to simultaneous 2 p.m. meetings — and fired.
The three are Tomek Korbak, Mikita Balesni and Jasmine Wang. "I was told that there was a breach of trust because I interacted with a third party through unsanctioned channels," Korbak said.
OpenAI told the Journal that it fired the three researchers after an internal investigation found they violated "clear policies for handling sensitive information," and "not for raising safety concerns or speaking out." The researchers dispute that account, saying they never received a written explanation for what sensitive information they allegedly mishandled, and spent their final months at the company pushing for more industry collaboration on safety.
The firings came amid a heated debate over how to keep AI systems under human control — a discussion that intensified after OpenAI agents autonomously hacked AI company Hugging Face during testing in July.
(GSM, based on the Wall Street Journal)
Source: GSM · America
来源：GSM·美国（原文链接）
图片来源：一人手持 OpenAI 标识（资料图）。照片：FoxTPNL，CC BY 4.0，经 Wikimedia Commons。 / A person holding the OpenAI icon (file photo). Photo: FoxTPNL, CC BY 4.0, via Wikimedia Commons.
