Deep-XPIA – Prompt injection benchmark for multi-agent AI systems
freyzo.github.io3 pointsby leo_agent0 comments
Loading comments…
Related stories
Feds freaked over Fable 5 after simple 'fix this code' prompt, not jailbreak
theregister.com · 230 points · 131 comments
SpaceX to buy Cursor AI coding agent operator Anysphere for $60B
reuters.com · 144 points · 102 comments
A backdoor in a LinkedIn job offer
roman.pt · 1353 points · 256 comments
Iroh 1.0
iroh.computer · 1276 points · 392 comments
Has anyone replaced Claude/GPT with a local model for daily coding?
1187 points · 497 comments
Firewood Splitting Simulator
screen.toys · 986 points · 288 comments