Red-teaming AI agent monitors with an LLM that fakes tool outputs

1 pointsposted 7 hours ago
by sgiannopoulos
(redwoodresearch.org)

2 Comments