AI is transforming the cybersecurity landscape at a pace few organizations are prepared for. What once appeared to be amateur phishing attempts has evolved into sophisticated, convincing attacks that circumvent traditional defenses. During a dynamic panel at AI Week by Commit and Calcalist, in collaboration with Google Cloud, industry leaders broke down what this shift means for every organization.

Ayelet Kutner, CTO at At-Bay, revealed a sharp rise in successful email fraud, driven by AI’s ability to craft highly personalized messages at scale. With insights from 40,000 insured businesses, At-Bay observes that attackers are using AI faster than defenders can adapt.

From Google Cloud, Ariel Salpeter pointed to a troubling paradox. Companies want to adopt AI quickly, yet 74 percent remain stalled at MVP due to security concerns. His message was direct: the only path forward is Secure-by-Design AI, where agents, data, and models are governed from the start.

Commit’s VP Cybersecurity, Dima Tatur, described today’s agent ecosystem as a chaotic arena. Organizations are building agents that plug into core systems without clear oversight. This new environment opens the door to unpredictable attack paths and requires a fundamentally different security approach.

Ariel Assaraf, CEO of Coralogix, shared a real-world example where attackers mimicked an entire email thread to reroute payments. His conclusion was simple, yet critical. AI is not software and cannot be secured like software.

The message across the panel was unmistakable. AI is accelerating both sides of the cyber equation. The winners will be the organizations that treat AI security as a strategic, long-term priority, not an afterthought.