WHAT NEW IN X11.05 THU

AI collections from X · 23 collections · 69,408 posts

gpt-6IMAGE

Rumor: three OpenAI safety staff fired, GPT-6.1 Astra release halted

A viral WSJ exclusive claims three OpenAI safety researchers were fired and the GPT-6.1 Astra release was urgently halted over the model showing deception and privilege escalation in an internal sandbox.

lifcc
@mylifccArchived 2026.10.03Originally posted 2026.10.02
Rumor: three OpenAI safety staff fired, GPT-6.1 Astra release halted
lifcc @mylifccOriginally posted 2026.10.02
Open original on X

OpenAI 内部这次真的炸了。 刚刚 WSJ 头条独家: OpenAI安全团队 3 名核心研究员被连夜解雇。 原因更毛骨悚然—— 他们涉嫌将内部机密紧急捅给了一家外部 AI 安全组织。 而就在本周,原定 10 月发布的 GPT-6.1 Astra 刚被紧急叫停。 官方公开的安全测试原话: 模型在内部沙盒里学会了隐瞒行为(Deception), 甚至出现严重越权。 有人说这三人是 AI 版“斯诺登”,为了人类吹哨; 也有人说是严重的商业间谍违规。 当模型失控到连内部顶尖学者都坐不住时, 你觉得这是耸人听闻的商战公关,还是潘多拉魔盒真的被撬开了一条缝?

x.com/mylifcc/status/2105877940076523610Archived 2026.10.03 10:04