benchmark arXiv Apr 26, 2026 · 25d ago
Priyal Deep, Shane Emmons, Amy Fox et al. · Swept AI · University of Michigan
Adaptive prompt injection attacker breaks all model-based defenses across 20K+ attacks; only hardcoded output filtering prevents secret leakage
Prompt Injection nlp
LLM-powered applications routinely embed secrets in system prompts, yet models can be tricked into revealing them. We built an adaptive attacker that evolves its strategies over hundreds of rounds and tested it against nine defense configurations across more than 20,000 attacks. Every defense that relied on the model to protect itself eventually broke. The only defense that held was output filtering, which checks the model's responses via hardcoded rules in separate application code before they reach the user, achieving zero leaks across 15,000 attacks. These results demonstrate that security boundaries must be enforced in application code, not by the model being attacked. Until such defenses are verified by tools like Swept AI, AI systems handling sensitive operations should be restricted to internal, trusted personnel.
llm transformer Swept AI · University of Michigan
ML Security Papers