Forescout Study Highlights How AI is Revolutionizing the Economics of Vulnerability Research

rifanmuazin - Monday, November 16, 2026 | 09:54
Finance Poetry
Forescout Study Highlights How AI is Revolutionizing the Economics of Vulnerability Research

The rapid advancement of artificial intelligence has long been framed as a double-edged sword for the cybersecurity industry. While security professionals fret over the potential for threat actors to weaponize generative models for more sophisticated phishing campaigns or automated malware development, a new report from Forescout Technologies offers a compelling look at how these same tools are fundamentally altering the defensive landscape. By leveraging off-the-shelf AI models, researchers have demonstrated that the traditional, time-intensive process of hunting for software vulnerabilities can be accelerated to an unprecedented degree, effectively lowering the barrier to entry for high-level security research.

In its newly released report, Finding Critical Infrastructure Bugs with Off-the-Shelf AI, Forescout details an experiment that challenges the conventional wisdom regarding the resources required to secure complex systems. The firm’s researchers sought to determine whether publicly available AI models could be harnessed to identify flaws in open-source software and proprietary binaries—tasks that have historically required deep expertise, specialized tooling, and months of dedicated human labor. The results, according to the firm, were transformative.

A Radical Shift in Efficiency

To put the theory to the test, the team at Forescout constructed a specialized AI-assisted vulnerability research harness. They then set this system against 10 widely used TCP/IP stacks, which serve as the foundational communication protocols for countless devices within critical infrastructure. The findings were stark: what would typically be a nine-month endeavor for a human research team was compressed into just 90 hours of total work.

During this condensed timeframe, the AI-assisted approach enabled the researchers to uncover 99 distinct vulnerabilities. Perhaps most concerning for industry stakeholders, 66 of these were classified as zero-day vulnerabilities—previously unknown flaws that could be exploited before a patch is developed. The breakdown of the process revealed that the vast majority of the time was spent on human oversight rather than raw computational analysis. The AI performed the heavy lifting in just 10 hours, while the subsequent 80 hours were dedicated to manual validation, ensuring that the findings were accurate and that the reported vulnerabilities were indeed genuine threats.

The economic implications of these findings are equally significant. Forescout reported that the static analysis conducted across the 10 TCP/IP stacks cost approximately $935. Broken down, this amounts to roughly $2.58 per vulnerability discovered. While the report notes that binary analysis proved to be more resource-intensive—costing about six times more and requiring twice the time compared to source-code analysis—the overall efficiency gain remains a paradigm shift for organizations operating on limited security budgets.

Daniel dos Santos, VP of Research at Forescout, emphasized that while the technology is powerful, it does not mark the end of the human expert’s role in cybersecurity. "AI did not eliminate the need for expert vulnerability researchers, but it dramatically compressed the work," dos Santos stated. "With access to an effective AI-assisted workflow, vulnerabilities in complex embedded software can be found much faster and at a fraction of the cost, by security researchers and attackers alike."

The New Economics of Cyber Risk

The implications of this research extend far beyond the laboratory. By changing the underlying economics of vulnerability discovery, AI is forcing a re-evaluation of how security teams identify, prioritize, and mitigate risk. For critical infrastructure sectors—where legacy systems, proprietary firmware, and hardware that is notoriously difficult to patch often remain in service for decades—the rapid identification of vulnerabilities is not merely a matter of efficiency; it is a matter of national and organizational security.

AI Cuts Vulnerability Discovery from Nine Months to 90 Hours

As AI tools become more accessible, the pace at which defenders must operate is accelerating. Forescout’s findings underscore a harsh reality: if defensive teams can use AI to find 99 vulnerabilities in a matter of days, malicious actors are undoubtedly using similar, if not more sophisticated, methods to do the same. This creates a high-stakes race between those seeking to harden these systems and those looking to exploit their inherent weaknesses. The legacy nature of much of the equipment used in industrial control systems and critical infrastructure makes this race particularly precarious, as many of these devices were never designed with modern, automated threat models in mind.

Navigating the AI Security Landscape

Despite the clear benefits demonstrated by Forescout, the industry remains cautious about viewing AI as a "silver bullet." The enthusiasm for AI-driven defense must be balanced against the reality of its limitations and the risks of over-reliance. Other industry research paints a more complex picture. For example, reports from ReliaQuest have highlighted that while AI can certainly make attacks cheaper and faster to scale, it also allows for a level of customization that makes threats significantly harder for traditional signature-based detection systems to identify.

However, the efficacy of AI in an offensive capacity is not universally agreed upon. Some security experts have expressed skepticism regarding the quality of AI-generated exploit code. A representative from Proofpoint recently noted that while AI can generate code at scale, the results are often highly flawed, lacking the precision and reliability that seasoned human hackers bring to the table. This "flawed output" problem suggests that while AI lowers the barrier to entry, it does not automatically turn an amateur into an elite cyber operator. It does, however, provide a force multiplier that, when paired with even a modicum of human expertise, can produce dangerous results.

This disparity in research underscores the necessity of a balanced approach. Security teams are increasingly being urged to adopt a "human-in-the-loop" philosophy, where AI handles the massive, repetitive task of scanning and initial identification, while human analysts focus on the high-level task of validation, risk assessment, and mitigation strategy. This hybrid model, as demonstrated by the 10-to-80-hour split in the Forescout study, appears to be the most effective way to harness the speed of machine intelligence without sacrificing the nuance and judgment required for effective cybersecurity.

For organizations looking to integrate these findings into their own security operations, the shift is not merely technological but cultural. It requires moving away from traditional, slow-moving manual audits toward a more continuous, AI-augmented vulnerability management lifecycle. As the cost of discovery continues to plummet, the traditional argument of "it is too expensive to audit every line of code" is losing its validity. In the coming years, the ability to rapidly identify vulnerabilities using automated, AI-driven research workflows may well become the baseline standard for any organization responsible for securing critical digital assets.

As the cybersecurity community continues to digest the impact of these findings, the message from Forescout is clear: the genie is out of the bottle. AI is already rewriting the rules of engagement in the digital arena. For security teams, the challenge lies not in fighting the tide of AI adoption, but in mastering these tools faster than those who seek to use them for harm. By embracing the efficiency of AI-assisted research, defenders can better protect the complex, often fragile systems that underpin modern society, ensuring that the next generation of security is as agile and scalable as the threats it faces.

Source : Finance Poetry
Author : rifanmuazin
Editor : Finance Poetry Editorial Team