Advanced AI tools help unskilled users initiate cyberattacks faster, according to RAND researchers. While these models accelerate early-stage reconnaissance, they fail to reliably execute complex, end-to-end offensive operations. This suggests a temporary gap between accessibility and actual capability. Security practitioners should prioritize defending against high-volume, low-skill probes rather than fully autonomous breaches.