| ▲ | simonw 9 days ago | |
The prompt injection stuff is very critical of both the technology and the LLM providers especially when I call out that their solution is still to say "they're getting better at avoiding the attacks" when my line has consistently been that "99% is a failing grade". | ||