With 31-Billion Daily Cyber Attacks, Global Defences Getting Maxed Out

Palo Alto Networks currently blocks about 31 billion cyber-attacks every 24 hours across its global footprint, Anand Oswal, executive vice president of network security at Palo Alto Networks, told ET. Not all of these are zero-day attacks, but a significant share involve previously unknown threats. The Santa Clara headquartered company also scanned around 4,000 open-source software packages over two months and identified 13,000-14,000 vulnerabilities. About 99.4% had not previously been disclosed, while about 40% were classified as high or critical severity. The number of vulnerabilities identified during that two-month exercise was comparable to what the company had found over roughly 13 years of operating its vulnerability research programme, pointing to the scale at which AI can speed up security research. The window available to organisations to stop cyberattacks is collapsing as AI agents become capable of finding vulnerabilities and executing attacks autonomously. Attacks that once required months of planning, and till more recently could be executed within days or minutes, can now unfold almost instantaneously once an AI agent identifies a viable path. Recent red-teaming exercises have shown that AI agents can repeatedly test systems, move beyond sandboxes and guardrails, locate vulnerabilities and identify ways to bypass protections. This widens the gap between attackers and defenders: attackers need to succeed only once, while security systems must stop every attempt. Oswal said enterprises need real-time protection across their infrastructure rather than relying solely on conventional detection and response.