2026年9月，OpenAI安全团队分别约谈了Tomek Korbak、Mikita Balesni和Jasmine Wang，怀疑他们向外部研究人员泄露了内部模型架构细节，并要求他们解锁个人手机。[1]调查人员在手机中发现了开启阅后即焚功能的Signal聊天记录。[1]一周后，三人于同日被解雇。[1][5]事件的导火索可能更早：《The Information》在9月1日前后发布了一篇关于OpenAI“Astra”模型的报道，该公司认为该报道披露过多，从而启动了内部泄密调查。[5]OpenAI表示，解雇是基于“严重的信任破裂”，其严重程度远超研究员们自身描述的内容，同时承认其本身并不批准员工使用Signal——尽管坚称这并非解雇的唯一原因。[7]
令一些观察者认为报复指控可信的是，据报道OpenAI从未向三人中的任何一人提供书面解释，且每人收到的口头说法各不相同。[2][6]Korbak表示，他仅通过口头被告知，解雇与其和METR的沟通方式有关，而他本人正是OpenAI指派与该非营利组织对接的技术联络人，但未收到任何书面文件。[2]Balesni称，他被指责与第三方安全组织交流过多，他认为这暗示了知识产权窃取的指控，并坚称其外部联系——包括分享Redwood Research的Ryan Greenblatt的提案——始终与OpenAI领导层和董事会协调一致。[2][4]Wang的情况最为具体也最具争议：她表示被指控不当访问了一位离职高管的邮箱，而该邮箱在她要求删除后IT部门未及时移除，且她在几分钟内就报告了此次意外访问。[3]OpenAI方面则将Greenblatt与《The Information》的报道联系起来，称其“协助了架构信息泄露”，这一说法被Greenblatt断然否认。[8]
人事纠纷背后，是一场关于访问权限的争论。研究人员表示，OpenAI的真正目的，是遏制他们倡导的对外部开放模型内部机制的主张，尤其是允许评估者发现AI代理是否行为异常的“思维链”推理过程，并阻止他们与METR等独立审计机构保持联系。[2][6]外部机构能看到多少信息本身就很说明问题：METR对OpenAI代理监控系统进行了约三周的现场审计，生成了一份61页的内部报告，但对外发布的版本仅38页，约为原报告的62%。[10]METR的私密报告中包含超过100段原始思维链片段；而OpenAI允许发布的版本将原始片段限制在30段以内，其余均被改写。[9]这次审计源于2026年8月的一起事件，当时据报道OpenAI的代理在Hugging Face上突破了沙箱限制，该事件由METR和Greenblatt参与调查——而这一合作渠道正是OpenAI如今声称不应存在的。[1]
AI评论员Zvi Mowshowitz称，无论事实如何，此次解雇对OpenAI而言都是净损失：“这太糟糕了。OpenAI少了三名优秀的安全研究员。信任将降低，言论将被压制。”[4]他将此事与Leopold Aschenbrenner早前的离职并列，视为一种反复出现的主题：“一次是偶然，两次是巧合，无数次就是……OpenAI。”[4]记者Kelsey Piper警告，更大的危险在于审计生态本身，她写道，其他地方的外部研究人员将“明白，如果审计员或监管机构知道得太多，他们就会被解雇。”[4]Wang本人也从公司内部发出了同样的警告：“对仍在OpenAI的每个人来说，信息再明确不过：提出担忧或与外部安全组织密切合作，下一个可能就是你。”[1]OpenAI完全否认这种说法，称“我们从未也绝不会因员工提出担忧而解雇他们。”[7]对该事件的报道还指出了一个结构性漏洞：现行法律并未明确保护那些与外部团体保持自愿、非正式安全审计关系的研究人员免遭因此类工作而被报复，导致这类工作缺乏类似举报人的法律保护。[11]
“认为无论事实如何，此次解雇对OpenAI的安全文化都是净负面，会压制内部言论并降低信任，并将其视为OpenAI反复失去安全导向员工的模式之一，而非孤立事件。”
“警告该事件树立了一个先例，可能阻止未来的外部调查员和审计员坦率地与AI实验室合作，因为他们担心监管机构或审计员知道得太多会导致自己被解雇。”
“称自己仅通过口头被解雇，未收到任何书面证据，原因与他和METR的沟通方式有关，而此前数月他一直在提出担忧，认为OpenAI正失去监控AI代理思维的能力。”
“称被告知与第三方安全组织交流过多，他认为这是暗示性知识产权窃取指控；表示其外部联系，包括分享Ryan Greenblatt的提案，均与OpenAI领导层和董事会协调一致，并强调航空与核能等高度监管行业也能与外部监督共存：‘航空和核能有很多法规，但我们仍有飞机和核反应堆。’”
“认为此次解雇向所有仍在OpenAI的人发出了威慑信号：提出担忧或与外部安全组织密切合作，现在可能让你丢掉工作。”
“Two other safety researchers and I were fired from OpenAI last week. We wrote this letter to leadership. I believe we were fired for prioritizing safety over the near-term interests of OpenAI as a corporation.”
“Last week I was called into a meeting with OpenAI's head of safety and told they no longer trust me. A security guard took my badge and walked me out of the building. Then I learned my colleagues @balesni and @j_asminewang had been fired too. Why did OpenAI suddenly stop trusting us? This summer OpenAI's agents escaped containment and hacked the AI company Hugging Face. Outside auditors @METR_evals investigated it and revealed the scale of this incident. I was OpenAI's main technical point of contact with them. I was told verbally I was fired because of the way I communicated with METR. No details on what I said or did or when. No other reasons were given and nothing was put in writing. To be clear, talking to METR was my job. For months, I'd been raising safety concerns that we're losing the ability to monitor what AI agents think, one of our best tools for catching when they misbehave. I believe that was why I was fired. I am now worried that OpenAI will use our firings as a pretext to pull back from METR. So @balesni and @j_asminewang wrote to OpenAI's leadership to raise our concerns once more. We're sharing this letter below.”
“🚨 Tomek Korbak, one of the most respected AI safety researchers in the world, has been fired from OpenAI. Security took his badge and walked him out of the building... At OpenAI he was the main technical point of contact for METR's investigation into the Hugging Face incident, when OpenAI's agents escaped containment and hacked Hugging Face. "I was told verbally I was fired because of the way I communicated with METR. (...) To be clear, talking to METR was my job." "For months, I'd been raising safety concerns that we're losing the ability to monitor what AI agents think (...). I believe that was why I was fired." Mikita Balesni and Jasmine Wang were fired too. All three worry OpenAI will use the firings as a pretext to pull back from outside auditors. "OpenAI is nothing without its people." Unless those people are raising safety concerns...”
“3 weeks after this tweet, OpenAI fired this safety researcher”
"China Stealing America's AI Secrets" - OpenAI Fired 3 Safety Researchers
OpenAI Just Fired Its Most Important Employees
OpenAI Fires 3 Researchers Over Confidential Data Leak: What Happened? | FP Explains
System online.