Why Frontier AI Models Are Creating A Vulnerability ’Storm’: Apiiro CEO

‘I'm not saying it as a prediction. I'm saying it with real, proven data from 95 percent of our customers,’ Apiiro CEO Idan Plotnik tells CRN.

The proficiency of frontier AI models at discovering software vulnerabilities and exploits is shaking up application security in what can only be described as a coming “major storm” for cybersecurity teams, according to Apiiro co-founder and CEO Idan Plotnik.

Yet, at the same time that frontier models such as Anthropic’s Claude Mythos are identifying flaws in large quantities—and at speeds much faster than traditional security tools—AI coding agents are also massively accelerating the rate at which new vulnerabilities are being created, Plotnik said.

[Related: How AI Agents Are Making Identity Security More ‘Critical’ Than Ever: Partners]

“It’s going to be a major storm," he told CRN. “I'm not saying it as a prediction. I'm saying it with real, proven data from 95 percent of our customers."

Organizations that have gained access to AI security models including Anthropic’s Claude Mythos and Fable, as well as OpenAI cyber models, have used them to scan their highest-impact applications and discovered hordes of vulnerabilities that other scanning tools had missed, according to Plotnik.

The surge in newly discovered vulnerabilities only makes prioritization more important than it was in the past, he said.

In addition to identifying and prioritizing vulnerabilities, Apiiro’s platform is also increasingly relying on AI agents to remediate the issues, Plotnik noted.

"We have the ability today to auto-fix any type of vulnerability that any of these frontier models will identify," he said.