Part 5 of a series of talks in which researcher Evan Hubinger explores the problems of safety for artificial general intelligence

This was recorded as part of the SERI ML Alignment Theory Scholars Program: https://serimats.org