Center for AI Safety

Having recently formed in April 2022, the organization is in its early developmental stages. Key activities during the reporting period include hosting the ML Safety Workshop and Trojan Detection Competition and launching its inaugural Philosophy Fellowship program. The ML Safety Workshop, held at the prestigious NeurIPS conference, drew over 140 papers and speakers from top universities, benefiting hundreds of attendees. The Trojan Detection Competition, done in collaboration with leading academic institutions, received over 1,200 submissions from 70 teams.

During the reporting period, the organization also launched its inaugural Philosophy Fellowship program which engages philosophers from various disciplines to contribute to the emerging field of AI safety through focused research and field-building efforts. Fellows critique and build on the existing conceptual AI safety literature, creating new conceptual frameworks to guide technical research, as well as produce original papers for publication.