Human cybersecurity mechanisms would be ineffective against a superintelligent AI, much like an ant's defenses against a human

The field of preventing harm from advanced AI systems, spanning alignment, evaluation, and decisions to slow or withhold models whose capabilities outpace their safeguards.

Artificial general intelligence: AI matching or surpassing humans across cognitive tasks.