Sara Price

Sara Price is a Member of Technical Staff at Anthropic working on AI safety, based in the Bay Area. She has worked in machine learning since 2016 and transitioned into AI alignment research in early 2024 through the MATS 5.0 (Spring 2024) program, after which she received a 6-month stipend to continue independent research on situational awareness and deception in AI systems. Her research focuses on adversarial robustness of multimodal LLMs, scheming and deception, control, and model organisms of misalignment. She was a co-author on Petri, an open-source AI safety auditing tool developed at Anthropic that automates evaluation of concerning model behaviors such as deception, sycophancy, and self-preservation. She now serves as a mentor for the MATS Summer 2026 program in the Anthropic and OpenAI Megastream.

No endorsements yet.

Personal Website

https://sbp354.github.io/

LinkedIn

LinkedIn

Twitter / X

Twitter

LessWrong

-

EA Forum

-