AI safety researcher

Guojie Song

Peking University

Google Scholar dblp ORCID Homepage

Last 12 months4 papers

AI safety papers from the last year

On this site: 5 papers

October 2026

LaVGuard: Selective Decoding for Contextual Value Drift
Oct 21arXiv ↗

September 2026

Values as Style: Disentangling Values from Semantics with One-Way Mixing for Low-Damage LLM Steering
Sep 30arXiv ↗

May 2026

Toward Stable Value Alignment: Introducing Independent Modules for Consistent Value Guidance
May 12arXiv ↗
Agent-ValueBench: A Comprehensive Benchmark for Evaluating Agent Values
May 11arXiv ↗

Authorship follows dblp and the author's own Google Scholar profile; only AI safety papers on this site are listed.