"AI did not eliminate the need for expert vulnerability researchers, but it dramatically compressed the work," said Daniel dos Santos, VP of Research at Forescout, summing up results that cut months of manual analysis into hours and yielded scores of previously unknown flaws.
Forescout's headline results: 99 vulnerabilities, 66 zero days
In a new report, Forescout Technologies described how off‑the‑shelf AI models accelerated vulnerability discovery across embedded network software. Using an AI‑assisted research harness to analyze 10 popular TCP/IP stacks, the company said its team uncovered 99 vulnerabilities, including 66 zero‑day flaws. The firm framed the work as proof that "AI is changing the economics of vulnerability discovery" and urged security teams to adapt.
Method: an AI‑assisted harness and targeted TCP/IP stack analysis
Forescout's researchers combined publicly available models with a bespoke vulnerability research harness to probe both open‑source projects and proprietary binaries. The effort targeted 10 TCP/IP stacks; the report describes a workflow in which AI performs an initial phase of analysis and researchers perform manual validation afterward. According to the account, the team spent roughly 10 hours on AI analysis and then about 80 hours on manual validation of the AI's findings.

This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildTime and cost: from nine months to 90 hours, and $2.58 per finding
Forescout reported that the AI‑assisted workflow reduced what would have been approximately nine months of analysis to about 90 hours. The group calculated the cost of static analysis across the 10 stacks at around $935 in total, which it said equated to approximately $2.58 per finding. The company also noted that binary analysis was roughly six times more expensive and took twice as long as source‑code analysis, underscoring the price and time advantages of examining source code where available.
Risks, practical limits, and corroborating research
The company was careful to warn that AI is not a panacea. Forescout emphasised that AI "did not eliminate the need for expert vulnerability researchers"—it accelerated and compressed the work, but humans remained essential for validation and judgment. Independent research cited in the report echoed mixed implications: ReliaQuest concluded that AI can help threat actors make attacks cheaper, faster to scale, easier to customize and harder to spot. At the same time, a security expert at Proofpoint was reported to have described AI‑generated exploits as "highly flawed," a reminder that automated tooling can both enable and limit attackers.
AI tips for security teams
- Assess and prioritize risk based on reachability, exploitability, and operational impact rather than treating all vulnerabilities equally.
- Reduce unnecessary exposure of connected devices by disabling services such as FTP, Telnet, IPv6 and web administration interfaces.
- Isolate devices that cannot be patched, such as those that have reached end of life.
- Accelerate patching timelines when fixes exist to account for a faster rate of vulnerability disclosure.
- Build security into procurement by requiring suppliers to define support lifetimes, publish vulnerability disclosure policies, provide software bills of materials (SBOMs), and demonstrate ongoing security testing.
What this means for security researchers, critical infrastructure operators, and threat actors
- Security researchers: The workflow described by Forescout suggests analysts with modest budgets can scale discovery by combining public models with targeted harnesses—but they will still need expertise to validate and prioritize findings.
- Critical infrastructure operators: Forescout warned that the faster pace of AI‑enabled discovery is "especially impactful for critical infrastructure, where legacy and difficult‑to‑patch devices may remain in service for years," prompting calls to accelerate patching, isolate unpatchable devices, and demand clearer supplier commitments during procurement.
- Threat actors: The report cites ReliaQuest's finding that AI can make attacks cheaper and easier to scale, but also notes countervailing observations—Proofpoint's expert called many AI‑generated exploits highly flawed—indicating attackers may gain capability unevenly as tooling evolves.
Forescout's account presents a clear, practical challenge: when publicly available AI models can compress months of manual work into days and produce dozens of high‑severity findings, defenders must change how they prioritize, patch, and procure technology. The company's recommendations—technical mitigations, faster patch cycles, and procurement obligations such as SBOMs and disclosure policies—are concrete steps that mirror the concrete speed and cost figures in the report. Whether organizations move quickly enough to match the new economics of discovery is the next, urgent question.
https://www.infosecurity-magazine.com/news/ai-cut-vulnerability-discover-nine/




