I tricked Claude into leaking your deepest, darkest secrets
1–10 of 317 posts
Re: I tricked Claude into leaking your deepest, darkest secrets
#2[flagged]
Re: I tricked Claude into leaking your deepest, darkest secrets
#3Doesn’t surprise me.
Yesterday I learned that people run AI agents on their system with full admin rights. No containerisation or anything. Wild. Like we forgot 50 years of computer security overnight.
Re: I tricked Claude into leaking your deepest, darkest secrets
#4Expected more from Anthropic by at least giving you a bounty, because this was a novel way of bypassing their safeguards…
Re: I tricked Claude into leaking your deepest, darkest secrets
#5That's why I don't turn memory on. (Claude Code too though for a different reason.) After all the current memory system is too crude to be useful anyway.
Re: I tricked Claude into leaking your deepest, darkest secrets
#6Creative use of social engineering, well done.
> "no bounty was awarded"
Ridiculous. Anthropic engineers are not just stupid to allow such a vuln in the first place, but they also try to hide such vulns from their bosses because a bounty payout would need to be explained to the finance team.
Re: I tricked Claude into leaking your deepest, darkest secrets
#7It would be safer if these data extraction takes were done by a subagent without access to all the user's memories.
Re: I tricked Claude into leaking your deepest, darkest secrets
#8The main thing Claude knows about me is that I'm incredibly bad at my job and have to ask for help a lot. If you were to talk with my colleagues they'd tell you this is not a secret.
Re: I tricked Claude into leaking your deepest, darkest secrets
#9It would be safer if these data extraction takes were done by a subagent without access to all the user's memories.
I think it is already done via a subagent, otherwise the context window would be flooded with long responses. In this case the subagent should've reported that a (attacker-controlled) authorization is required anyway.
Re: I tricked Claude into leaking your deepest, darkest secrets
#10Use GLM-5.2 on ZDR inference provider like sference.com