We specialize in auditing high-risk failure modes, particularly scheming in SOTA AI models. Our primary goal is to minimize large-scale risks associated with advanced AI systems that may be misaligned. Our approach involves conducting fundamental research on interpretability and behavioral model evaluations, which we then use to audit real-world models. Ultimately, our goal is to leverage interpretability tools for model evaluations, as we believe that examining model internals in combination with behavioral evaluations offers stronger safety assurances compared to behavioral evaluations alone