IBM: New Cybersecurity Measures Against AI Agent-Driven Attacks

IBM today announced two new cybersecurity solutions designed for an era in which attackers use frontier AI models to automate attacks on enterprises.

What Are Agentic Attacks

According to IBM, agentic attacks represent a new class of threats where attackers use advanced AI models to:

• Autonomously discover vulnerabilities in complex IT environments
• Accelerate all attack phases — from reconnaissance to exfiltration
• Operate at machine speed, creating continuous business disruption
• Reduce the time, cost, and expertise needed for sophisticated attacks

Two New Solutions

1. Enterprise Cybersecurity Assessment — an assessment service conducted by IBM Consulting. It identifies security gaps, AI-specific exposures, and provides prioritized mitigation recommendations.

2. IBM Autonomous Security — a multi-component service that uses AI agents for coordinated threat response. It includes automated vulnerability analysis, policy enforcement, anomaly detection, and threat containment with minimal human intervention. Key feature: it works independently of the security tool vendor.

Why It Matters

While the industry focuses on building AI agents, IBM warns about the other side of the coin — attackers are already using the same capabilities. Enterprises need new defenses designed for threats that move at AI speed, not human speed.