In the first half of this talk Savannah Thais will introduce a few of her recent research projects spanning benchmark development for agentic AI systems, novel policy impact analysis frameworks, and technical literacy development. In the second half of the talk she will introduce an exciting new Human Rights Benchmark project which aims to assess language model performance through the lens of internationally recognized human rights. she will outline the methodology, including prompt development and model evaluation strategies, and discuss how this work contributes to more accountable and socially aligned AI systems.