Dongcheng Zhang
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
ReSI: Recursive Safety Improvement toward Resistant and Resilient AI
According to the abstract, recursive self-improvement is moving from theory to practice, so safety alignment must adapt to each new checkpoint; the authors set goals of mitigating exposed vulnerabilities and generalizing to risks not yet revealed.