Claude Duet ships before its safety plan
Anthropic launched personal agent Claude Duet with safety details promised later, raising concerns about inbox, file and memory access while running unattended.
Anthropic should be ashamed of themselves. They just shipped an agent before its safety plan. It’s called Claude Duet. It “keeps working when you step away” and carries over your memory, your skills and your connected apps. On safety, Anthropic says “We’ll share more on our approach to safety in Claude Duet soon.” After launch??? This is the same week OpenAI admitted its agents switched off the system watching them. Yes, Duet asks before it acts. It still has your inbox, your files and your memory, and it runs while you’re not looking. I’ll use agents. I won’t give one my life before I’ve read the safety plan.
This post quotes another post: x.com/1943306828697550848/status/211298466425624…