llama3.2

I ran 500 automated attacks against an on-prem AI. The grandma attack passed nearly half the time. A few weeks ago I published the results of a manual red team...