AI now runs parts of cyber attacks, Check Point says
What happened
Check Point's 2026 AI Security Report documents attackers using AI to perform steps of live intrusions, including turning disclosed vulnerabilities into working exploits within hours. The report gives detailed examples of AI executing thousands of commands in multi-session campaigns, making shorter patch-and-response windows operationally real. Watch whether defenders adopt mandatory telemetry and faster supplier-forensic commitments in contracts
Why the category manager should care
Treat AI-driven attacks as an operational reality and require evidence-grade telemetry and forensic cooperation in supplier contracts
Key facts
- AI can convert a newly disclosed vulnerability into a working exploit within hours
- Example case: AI executed thousands of commands across dozens of attack sessions
- Detections of long malicious prompt-injection payloads rose about fivefold between March and