The potentially negative impact of Frontier AI threats cannot be underestimated by CISOs, C-suite executives, and board members. This is driving organizations to radically rethink their security strategies before Frontier-discovered vulnerabilities can be exploited and create massive damage.
This is an acutely strategic issue because of the previously unimaginable speed and scale of Frontier AI threats in discovering and exploiting vulnerabilities, putting unprecedented pressure on cyber defenders to identify and plug gaps faster than ever. The recent discovery of critical vulnerabilities in the PHP programming language made it clear hackers can swiftly compromise huge numbers of web servers and web sites.1 It also is clear that the PHP example will undoubtedly be followed by similar attempts due to AI-driven discovery of vulnerabilities.2
Research from Palo Alto Networks’ Unit 42 threat intelligence arm puts this in perspective. Proactive vulnerability discovery by their Network and Open Source Vulnerability Analyzer (NOVA) over a 3-month period indicates that more than 20,000 zero-day threats were identified, with a startling 99.4% of those findings having had no public Common Vulnerabilities and Exposures (CVE) record.3
The speed and scale of Frontier AI vulnerabilities exposure has driven cybersecurity leaders to develop a new mindset for cybersecurity protection—one that replaces reactive, event-driven steps with a more proactive approach shaped by new tools, services, and models. This is essential because traditional cybersecurity frameworks’ threat detection and prevention approaches have been built upon the framework of multiple, disparate point products; this outdated cyber process risks being shredded by hackers whose capabilities will be dramatically enhanced by the existence of Frontier AI models.
The degree to which mission-critical operations may be compromised by Frontier AI risks is alarmingly high: The previously mentioned NOVA analysis indicated that nearly 40% of those zero-day threats carried either a High or Critical severity ratings.
New strategies for new threats
It's time for new ideas and new strategies, because the potential impact on business operations and sustained resiliency is more ominous than ever. Importantly, this no longer is solely a CISO-centric solutions model, but instead represents a comprehensive, cross-functional leadership initiative. It requires an updated strategy model that spans the full spectrum of environments, infrastructure, and cybersecurity tools and services. Just as AI is being used—aggressively and successfully—to identify and exploit vulnerabilities faster and more efficiently, AI also must be used to drive this new threat management strategy.
For instance, organizations should incorporate new AI tools into their strategies for mitigating this rapidly expanding vulnerabilities landscape. In fact, the same Frontier AI models that represent a potential challenge to cybersecurity resilience should be used to identify and block threats using proactive vulnerability discovery. An important source of this ability to identify and patch security vulnerabilities and gaps even before formal patches are available is Palo Alto Networks’ Frontier Virtual Patching, which represents an innovative breakthrough in pre-emptive protection.4
Identifying and blocking Frontier-based threats before they do damage
Using Frontier AI in concert with its own Precision AI technology, Frontier Virtual Patching provides cloud-based security services that efficiently detect and prevent rapidly emerging threats. Using AI-powered security models, it analyzes, identifies, and issues virtual patches for malicious threats even before formal security patches have been issued by software vendors, helping organizations get out in front of new threats before they can be exploited. It also provides fine-grained, high-fidelity data to continuously train its security models for comprehensive insights into new and still-emerging threats.
Frontier Virtual Patching pays off on organizations’ new strategic efforts by providing a proactive vulnerability discovery and pre-emptive protection model before those risks are weaponized. The NOVA open source discovery system pulls together multiple leading Frontier AI models into proprietary harnesses to scan thousands of open source software repositories, helping to rapidly identify hidden weaknesses.
It's time to rethink cyber strategy for the Frontier AI era
Frontier AI models are both exciting in their defensive capabilities and frightening in their potential for widescale damage. Threats and vulnerabilities can no longer routinely and efficiently be addressed at human-speed timelines. Traditional linear solutions models—identify the vulnerability, wait for the patch, deploy, and adjust—simply don’t work with Frontier AI models.
AI-powered attacks are increasing pressure on cyber defenses; the good news, however, is that more and more organizations understand and acknowledge that reality. In fact, nearly half of all organizations (44%) have high expectations in the use of new AI solutions to defend against Frontier AI.5
In order to actually reap the benefits of AI-based solutions (such as Frontier Virtual Patching) against Frontier AI risks, organizational leadership needs to start with a clean sheet of paper. Consider the following issues to be essential elements in an organization’s Frontier AI defense strategy:
- Zero Trust architecture and identity security should be front-and-center to improve access controls and enforce least-privilege policies. This must account for non-human identities (such as agents) as well as human identities.
- Proactive vulnerability management and virtual patching helps identify, prioritize, and mitigate vulnerabilities before exploits occur.
- Use AI to defend against AI by leveraging automation and real-time intelligence to spot and block increasingly complex and persistent attacks.
Be sure to check out the second in this two-part series on Frontier AI risks and solutions, “Cybersecurity Patching Needs a New Model.”
1 AI tools discovered the PHP vulnerability that could take down 70% of the internet, Fortune, October 2026
2 Silicon Valley is grappling with a flood of bugs found by AI, New York Times, October 2026
3 Harnessing AI for prevention: Introducing Frontier Virtual Patching with PAN-OS 12.2, Palo Alto Networks, August 2026
4 Protecting vulnerable systems by using Frontier Virtual Patching: Solutions guide, Palo Alto Networks, October 2026
5 State of critical infrastructure cybersecurity report, Palo Alto Networks, 2026