AIChilles Risk Discovery · Weakness Explorer

Weaknesses discovered in AI-evolved systems.

Real results from the weakness-discovery pipeline: filter by app, evolution framework, and LLM, then inspect each root-cause cluster — the evolved code, the workload that triggers it, the LLM's root-cause hypothesis, and a P-vs-P′ regression curve built from the discovered witnesses.

App
AI-evolve framework
LLM

Discovered weaknesses

Real discovered witnesses grouped by root cause.

0
Witnesses
0
Correctness
0
Regressions
0
Root causes
Choose a weakness to inspect its trigger workload and evidence.