Why AI Is Accelerating Old Cyber Risks, Not Creating New Ones
TechTarget, Tuesday, August 19th, 2025
AI changes the speed and scale of exploitation rather than introducing genuinely new categories of risk.
LevelBlue pentesting head Ed Williams argues the underlying vulnerabilities organizations face today are the same ones from five or ten years ago: unpatched systems, weak identity controls, excessive privileges, and insecure third-party integrations.
AI changes the speed and scale at which those weaknesses are found and exploited, automating reconnaissance and lateral movement and making malware iteration, deepfakes, and phishing easier to scale.
Treating AI as an exceptional threat creates strategic distraction, pushing teams toward AI-specific tools while patch management, asset inventories, and identity governance remain the most reliable entry points. Prompt injection is best understood as an extension of input validation and data integrity concerns.