Is it already too late to reduce catastrophic AI risk?

New Kennedy School ‘pre-mortem’ lays out threats, vulnerabilities, and possible steps, warns firms must act now