3 ms·Catastrophic sabotage as a major threat model for human-level AI systems5 points by speckx 2y ago