Back to Paper
cs.CRcs.AI
Local ID: 2604.23887v2
AI Summary: gemma4:e4b
Evaluation of Prompt Injection Defenses in Large Language Models
By Priyal Deep, Shane Emmons, Amy Fox, Kyle Bacon, Kelley McAllister, Peter Ortiz, Krisztian Flautner
Revision History Timeline
v14/26/2026
4/26/2026
“14 pages, 9 figures”
v25/12/2026
5/12/2026
“14 pages, 9 figures”
★ Version indexed in ExplorerComparing v1 vs v2
Green = Added • Red = Removed
Title Comparison
Evaluation of Prompt Injection Defenses in Large Language Models
Authors Comparison
Added:Peter Ortiz
Unchanged:Priyal Deep, Shane Emmons, Amy Fox, Kyle Bacon, Kelley McAllister, Krisztian Flautner
v1 Comment
“14 pages, 9 figures”
v2 Comment
“14 pages, 9 figures”
Abstract Word Diff
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.