Affiliate Bruce Schneier and coauthors argue that prompt injection attacks are the first step of a seven-step promptware kill chain. The seven steps they propose are: initial access, privilege escalation, reconnaissance, persistence, command-and-control, lateral movement, and actions on objection. The authors propose "an in-depth defensive strategy that assumes initial access will occur and focuses on breaking the chain at subsequent steps, including by limiting privilege escalation, constraining reconnaissance, preventing persistence, disrupting C2, and restricting the actions an agent is permitted to take."
Read or listen to the authors' piece from Lawfare.
You might also like
- communityGabon blocked social media
- communityHow sky-high AI pay warps science
