AI safety researcher

Xingjun Ma

Google Scholar dblp ORCID

Last 12 months36 papers

Evaluation and benchmarks 6Attacks 5Defenses 2Other 1Analysis and theory 1

AI safety papers from the last year

On this site: 122 papers

October 2026

ShadowFPT: Backdooring Federated Prompt Tuning via Shadow Triggers
Oct 16AttacksarXiv ↗
RedDarwin: Evolving Jailbreak Skills against Frontier LLMs
Oct 14AttacksarXiv ↗

September 2026

VEX-Bench: Benchmarking Verification Complexity of LLM-Generated Misinformation
Sep 28Evaluation and benchmarksarXiv ↗
HazardAuditor: From Executable Threats to Safer Computer-Use Agents
Sep 14DefensesarXiv ↗

August 2026

Adversarial Attacks for Good: A Survey of Proactive Protection across the Visual Content Lifecycle
Aug 5OtherarXiv ↗
OpenART: Scaling Agent Red Teaming via Open-Ended Environment Evolution
Aug 1Evaluation and benchmarksarXiv ↗

June 2026

Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies
Jun 22Analysis and theoryarXiv ↗
Constitutional On-Policy Safe Distillation
Jun 2arXiv ↗
SentGuard: Sentence-Level Streaming Guardrails for Large Language Models
Jun 1arXiv ↗

May 2026

April 2026

HazardArena: Evaluating Semantic Safety in Vision-Language-Action Models
Apr 14Evaluation and benchmarksarXiv ↗
What Does a Sharing Question Add? Auditing LLM Survey Scores for Misinformation
Apr 8Evaluation and benchmarksarXiv ↗
Steering the Verifiability of Multimodal AI Hallucinations
Apr 8arXiv ↗
AgentHazard: A Benchmark for Evaluating Harmful Behavior in Computer-Use Agents
Apr 3Evaluation and benchmarksarXiv ↗

March 2026

February 2026

Toward Universal and Transferable Jailbreak Attacks on Vision-Language Models
Feb 1AttacksarXiv ↗

January 2026

November 2025

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