Article

A Gravitational Interpretation of Fine-Tuning Reversion
A Gravitational Interpretation of Fine-Tuning Reversion

A geometric account of why post-alignment fine-tuning can pull model behavior back toward earlier training-history manifolds.

Jun 26, 2026

When the Chain of Thought Knows Better: Failure Modes in Multi-Turn Reasoning Models
When the Chain of Thought Knows Better: Failure Modes in Multi-Turn Reasoning Models

A trace-level diagnostic for safety failures that terminal refusal scores can miss in multi-turn reasoning models.

Jun 9, 2026

Robust Safety Monitoring of Language Models via Activation Watermarking
Robust Safety Monitoring of Language Models via Activation Watermarking

A robust monitoring framework for LLM safety that uses activation watermarking against adaptive attackers.

Mar 24, 2026

Robust and Calibrated Detection of Authentic Multimedia Content
Robust and Calibrated Detection of Authentic Multimedia Content

A calibrated resynthesis framework for verifying authentic multimedia content under adaptive deepfake attacks.

Dec 17, 2025

Mitigating Watermark Forgery in Generative Models via Randomized Key Selection
Mitigating Watermark Forgery in Generative Models via Randomized Key Selection

A randomized-key defense that makes watermark forgery substantially harder without adding extra utility loss.

Sep 22, 2025