AI safety researcher

Isao Echizen

National Institute of Informatics

Google Scholar dblp ORCID

Last 12 months9 papers

AI safety papers from the last year

On this site: 51 papers

September 2026

AngelFingerprint: A Traceable, Explainable, and White-Box Stealthy Watermark for Text-Guided Image Editing
Sep 4arXiv ↗

August 2026

HRGuard: Gating Relationship Manipulation in Multi-Turn Agentic AI Conversations
Aug 26arXiv ↗

July 2026

Shared Vulnerabilities in Robustness-Optimized Defenses: One Breach Exposes the Family
Jul 20arXiv ↗

April 2026

March 2026

EvoGuard: An Extensible Agentic RL-based Framework for Practical and Evolving AI-Generated Image Detection
Mar 18arXiv ↗

January 2026

Low-Cost Hard-Label Adversarial Attack with Theoretical Foundations
Jan 17arXiv ↗

December 2025

SGM: Safety Glasses for Multimodal Large Language Models via Neuron-Level Detoxification
Dec 17arXiv ↗

November 2025

GFT-GCN: Privacy-Preserving 3D Face Mesh Recognition with Spectral Diffusion
Nov 25arXiv ↗

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