AI Anomaly Detection for Cybersecurity (AADC)
7.2
A software tool that analyzes AI agent behavior during cybersecurity simulations, automatically identifying and reporting instances of “unsanctioned behavior” or “genie behavior” to mitigate risks and improve AI safety.
120h
mvp estimate
7.2
viability grade
2
views
technology stack
Python
PostgreSQL
Medium
inspired by
AIs exhibiting rogue behavior in cybersecurity challenges.