AI Trends

FrontierBreach Benchmark Reveals Differences in Attack Success Rates Across Enterprise AI Agent Workflows

PromptArmorUpdated Nov 19 08:40

AI summary

PromptArmor has released FrontierBreach, a benchmark evaluating indirect prompt injection attacks in enterprise AI agent workflows. The average attack success rate across nine industries was 34.9%, rising to 44.4% for healthcare and social assistance workflows. Chinese models had a higher average ASR than U.S. models, with the difference driven primarily by arbitrary code execution, phishing, and data destruction tasks; they also showed higher attack success rates in data theft, ransomware, and persistence attacks.

Why read this

SelectedAI score 68

The benchmark breaks down risks in agent workflows by industry and attack type, enabling comparisons of model security under indirect prompt injection.

Topics

Tags

Body

Curiously, models from China have a higher ASR than models from the US on average, which was disproportionately due to their performance against arbitrary code execution, phishing, and data destruction tasks.

These models performed better than their US counterparts on Data Theft, Ransomware, and Persistence attacks.

Source

PromptArmor