Gates says AI crossed five danger thresholds

Bill Gates says frontier AI has crossed five danger thresholds—bio-capability, cyber-capability, psychosocial capability, job-market disruption, and loss of control—and argues that any model able to make novel molecules should be monitored, according to his interview with MIT Technology Review. Gates puts bioterrorism risk at roughly 50 times the risk of a natural pandemic, a personal estimate rather than a measured forecast; builders should file the warning as a governance prompt to define capability triggers and escalation owners, alongside the archive’s analysis of the entry-level employment gap.