Universal Adversarial Suffixes Against Tool-Calling LLMs
Adversarial suffixes hijack tool-calling models to execute unauthorized actions on live systems.
Priya Subramaniam
Section
Adversarial ML and Prompt Attacks
2 stories in Adversarial ML and Prompt Attacks.
Benchmarking Prompt Injection Defenses in Production Agents
Existing benchmarks miss production agents' full attack surface and cascading failures.
Madeleine Truchon