The majority of attacks that are happening right now are not because attackers are finding more vulnerabilities, but because they are exploiting more existing vulnerabilities.
The vast majority of modern security breaches stem from familiar, unresolved entry points such as compromised credentials, outdated systems, and phishing campaigns. In a conversation with iTnews Asia, Check Point Chief Technology Officer Jonathan Zanger said that threat actors rely on established weaknesses to move through enterprise networks more quickly.
“The majority of attacks that are happening right now are not because attackers are finding more vulnerabilities. It’s because they’re exploiting more existing vulnerabilities,” Zanger told iTnews Asia.
Data from Microsoft Security Blog analyses covering CVEs published through September 2026 show that approximately 383,000 software vulnerabilities were disclosed across platforms and technologies from 1999 through September 2026. However, fewer than 1 in 200 published CVEs—about 2,000 in total—have ever been linked to real-world exploitation in the wild.
Attackers disproportionately target flaws that grant system takeover capabilities.
Remediation timelines vary significantly across enterprise environments. Microsoft Security Exposure Management data shows that median vulnerabilities are patched in roughly two weeks on Windows workstations, but take more than a month on Windows servers, over 10 weeks on Linux servers and macOS applications, and more than a year on network appliances.
As organizations expand their use of artificial intelligence, cybersecurity teams face the challenge of accelerated attacks alongside older security gaps. Zanger recommended that enterprises improve patch management, tighten network segmentation, and utilize existing threat prevention tools rather than replacing infrastructure.
Addressing the deployment of autonomous AI agents that act on behalf of employees, Zanger emphasized the need for architectural foresight. “Design the system initially with security in mind, so you wouldn’t have the discussion in a couple of years of how to make it secure,” he said.
He added that traditional security tools cannot manage agents operating across multiple enterprise systems independently. “Organisations need an external security mechanism to monitor an agent’s purpose, permissions and actions, and block activity that falls outside its authorised scope,” Zanger said.
“Really understanding the long tail gaps is a difficult problem.” “And if we can leverage AI in that, I really think that’s the most effective way to improve your security posture,” he said.