Why do AI labs that warn about dangerous AI keep so much in Slack?
Why do AI labs that warn about dangerous AI keep so much in Slack?
Cybersecurity startup Cathedral reached Anthropic’s sales, HR and IT systems, including Slack messages, by exploiting third-party software.
Anthropic says the attempts took several days and that source code, model weights and Claude chats weren’t exposed. Former OpenAI researcher Will Depue says Anthropic is downplaying it.
Should AI labs meet a higher security standard?
Watch the full episode of In The Loop on our channel.
