Tags
13 pages
Agents
economics of open versus closed weights
gradient-free jailbreaks and cpu-side suffix search
active defense and adversarial agents: coordination poison (exploratory idea)
active defense and adversarial agents: GCG attacks
active defense and adversarial agents: service sandbagging
active defense and adversarial agents: context bombs
active defense and adversarial agents: exploding search space (academic)
active defense and adversarial agents: malicious skills
active defense and adversarial agents: hostname beacon
active defense and adversarial agents: token burn
1
2