overview
What is Center for AI Safety (CAIS)?
Center for AI Safety (CAIS) is a non-profit organization focused on AI safety developed by Center for AI Safety (CAIS) that enables researchers, engineers, policymakers, and the public to conduct research, build the field, and advocate for safety standards to reduce societal-scale risks from AI. It was founded in 2022 by Dan Hendrycks and Oliver Zhang.
CAIS is a San Francisco-based non-profit organization dedicated to mitigating high-consequence, societal-scale risks posed by artificial intelligence, which it categorizes alongside global priorities such as pandemics and nuclear war. The organization operates through three primary pillars: advancing safety research, field-building, and advocacy.
Its research initiatives encompass both technical and conceptual approaches. Technical research focuses on developing foundational benchmarks and methodologies to identify and address AI safety issues, including the removal of dangerous behaviors, the study of deceptive AI, the training of AIs for moral conduct, and the improvement of system reliability and security. Conceptual research examines AI safety from multidisciplinary perspectives, integrating insights from safety engineering, complex systems theory, international relations, and philosophy to construct frameworks for understanding challenges and societal risks.
For field-building, CAIS aims to expand the AI safety research ecosystem by providing funding, research infrastructure, and educational resources. This includes offering free access to its compute cluster for qualified researchers, administering multidisciplinary fellowships such as the AI and Society Fellowship and the Philosophy Fellowship, and developing educational content like the "Intro to ML Safety" course.
In its advocacy role, CAIS advises industry leaders, policymakers, and governmental bodies to elevate public awareness of AI risks, inform policy development, and establish guidelines for the safe and responsible deployment of AI. A significant advocacy effort was the May 2023 statement on AI risk of extinction, which garnered signatures from over 1,000 AI professors, industry leaders including Sam Altman and Elon Musk, and public figures.