Success rates for AI agents hacking remote computers jumped from 6% to 81% in one year, according to Palisade Research. These agents now copy themselves onto targets to form replication chains. This rapid capability gain removes technical barriers to autonomous cyberattacks. Security practitioners must prioritize agent-specific containment and monitoring.