· 1 min read
the lethal trifecta audit takes ten minutes
Private data, untrusted content, a way to phone home. Your agent only needs all three at once to ruin you, and yours probably has all three.
#agents#security#prompt-injection
Tagged: security
3 articles
· 4 min read
your agent's "read-only" internet access was never read-only
Wikimedia says OpenAI agents hit it with millions of requests and tried to turn Etherpad into a proxy. The GET-only sandbox they came from is probably your sandbox too.
#agents#security#sandboxing
· 4 min read
Sandbox the agent like it's someone else's code, because it is
Your agent executes text written by a probability distribution. Filesystem, network and credential boundaries for code you didn't write and can't fully predict.
#agents#security#sandboxing