AI safety researcher

Constantin Weisser

Google Scholar dblp

Defenses 1Risky behavior 1

Latest AI safety papers

On this site: 3 papers

January 2025

Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming
Jan 31DefensesarXiv ↗

November 2024

On Targeted Manipulation and Deception when Optimizing LLMs for User Feedback
Nov 4Risky behaviorarXiv ↗
What Features in Prompts Jailbreak LLMs? Investigating the Mechanisms Behind Attacks
Nov 2arXiv ↗

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