Henry Sleight

All Code, No Thought: Language Models Struggle to Reason in Ciphered Language

The Hot Mess of AI: How Does Misalignment Scale With Model Intelligence and Task Complexity?

Beyond Data Filtering: Knowledge Localization for Capability Removal in LLMs

Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignment

Unsupervised Elicitation of Language Models

Abstractive Red-Teaming of Language Model Character

Persona Vectors: Monitoring and Controlling Character Traits in Language Models

The LLM Has Left The Chat: Evidence of Bail Preferences in Large Language Models

Adaptive Deployment of Untrusted LLMs Reduces Distributed Threats

Best-of-N Jailbreaking

Looking Inward: Language Models Can Learn About Themselves by Introspection

Failures to Find Transferable Image Jailbreaks Between Vision-Language Models

Quantifying Elicitation of Latent Capabilities in Language Models

Rapid Response: Mitigating LLM Jailbreaks With A Few Examples

Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs

Plan B: Training LLMs to fail less severely

Attacking Audio Language Models with Best-of-N Jailbreaking