Question: Constitutional AI and RLHF Alignment Comparison

Red-teaming attack success rate on CAI models

In audit

0/5 publication-ready

Findings metrics

—

total

—

verified

—

pending

—

rejected

Research