Israeli experts are warning that AI is exponentially increasing cyber threats

SEPTEMBER 14, 2026 10:37Updated: SEPTEMBER 14, 2026 10:39A global coalition of more than 150 cybersecurity companies has warned that artificial intelligence is rapidly increasing the speed and scale of cyberattacks, and called on governments and businesses to update their defenses before attackers gain a lasting advantage.Among the signatories are major industry players, including Anthropic, OpenAI, KPMG, Visa, Oracle, Microsoft, and Palo Alto. They called on organizations to “raise the security bar” and for governments to immediately provide and coordinate defensive cyber technology at the local, national, and international levels. The coalition’s statement said that AI systems now allow attackers to identify and exploit vulnerabilities far faster than human defenders can respond.“In the coming months, AI-enabled cyberattacks will become far more widespread and sophisticated as models around the world become increasingly capable. The companies and public services our communities depend on - from hospitals to water treatment plants to the infrastructure that powers the internet - are at risk,” the letter said.Anthropic, the developer of the artificial intelligence model Claude.AI (credit: DADO RUVIC/REUTERS, ILLUSTRATION)KPMG’s head of cybersecurity in Israel, Omer Or, told The Jerusalem Post that organizations are entering a period where traditional security models may no longer be sufficient. Or, who has spent more than two decades in senior government roles, described the current shift as unprecedented, with AI now boosting offensive cyber capabilities by all levels of threat actors. “We are already in a rapid technological race, and AI is moving very quickly,” he said. “If we look at offensive cyber of all levels, from the individual hacker to criminal cartels and state actors, all are in the process of getting a boost to their offensive capabilities and have access to tools that were once only available to state actors. That creates a change in the equation.”According to Or, one of the most significant shifts involves zero‑day vulnerabilities, which were previously the domain of highly resourced state actors.“With these frontier models, we see the pace at which attackers can produce zero‑day vulnerabilities, 100 times more than humans can do on their own,” Or said, adding that AI is now finding weaknesses in legacy and operational‑technology systems that have historically been overlooked. “These areas received less attention and now it’s becoming identical.”Old problems, new speedWhile some imagine AI will introduce exotic new attack types, Ariel Litmanovich, founder and CTO of Aryon Security, told the Post that the opposite is true. “Most people imagine that the main threat would be new futuristic attacks, but actually AI makes the weaknesses much easier to find and exploit. And that’s what the letter says.”Litmanovich, who previously worked on the IDF’s transition to public cloud environments, said the vulnerabilities attackers are exploiting today, including misconfigurations, excessive permissions, and weak authentication, are the same ones that have plagued organizations for years. The only thing that has changed is the pace.In the past, organizations could afford slow remediation cycles. A cloud resource might be deployed insecurely, detected later by a security tool, and fixed eventually. That model, Litmanovich said, no longer works.“It’s the same problems; it’s just more dangerous because defenders don’t have the time to fix them. Time to exploit has gone down disproportionately because of AI,” he said. “Attackers can increasingly scan, analyze, adapt, and exploit the vulnerabilities or security issues at a speed and scale that humans cannot match manually. If defenders want to be ahead, they need to find a way to be faster than the attacker.”To do so would require integrating AI into an organization’s internal defensive operations and quickly identifying anomalies. Otherwise, attackers will continue to exploit vulnerabilities and defenders won’t have enough time to remediate and fix the problems.“Cybersecurity needs to move towards containment, resilience, and a quick incident response for business continuity,” he said. “Organizations have to be more proactive and preventative to stay ahead and break the status quo. You can’t fight machines and AI with humans. Humans can’t win against AI.”Follow us on Google