Chubby 就 Will Depue 关于 Anthropic Slack 工作区被取得完整管理员权限、可冒充员工的爆料作出评论，认为若属实潜在损害会很严重。他指出训练方法、未发表结果、失败实验和未来模型计划可能暴露实验室的有效做法与方向，让竞争对手省去试错成本；冒充员工则可能让恶意请求伪装成同事的日常消息，打开进入其他系统的路径。
If Will Depue’s claims of full Slack access and employee impersonation at Anthropic are accurate, the potential damage would be severe.
Training methods, unpublished results, failed experiments and plans for future models could reveal what works, what doesn’t, and where the lab is heading. That knowledge could spare competitors expensive trial and error, even without access to model weights.
The ability to impersonate employees would create another risk: malicious requests could look like routine messages from trusted colleagues, potentially opening paths into other systems.
So yeah, if this turns out to be true, this is one of the biggest threads Anthropic will face to date.
anthropic is massively underplaying the recent security breach. i'm hearing cathedral achieved full administrator access of the slack workspace (see every message, log in as arbitrary users) where literally every single research secret is contained to those unfamiliar: Ant & OAI keep literally everything in Slack. it sounds exactly the same incident that happened to OpenAI in 2023, where OpenAI was hacked by a random guy who got access to an access token from an external app service "we're critical to national security, we must beat China, and yes we keep all our crown jewels in Salesforce's Slack messenger. yes we have admin API tokens everywhere. why do you ask?"在 X 查看被引用的帖子
来源：Chubby♨️ · x.com