Infosecurity
Forescout uses AI to find 99 flaws across TCP/IP stacks
Forescout Technologies has highlighted how AI can dramatically shorten vulnerability discovery timelines after releasing a report detailing an AI-assisted approach that worked on 10 popular TCP/IP stacks. The researchers claim that using publicly available models with a custom vulnerability research harness reduced analysis time from about nine months to 90 hours, yielding 99 vulnerabilities, including 66 zero days.
After 10 hours of AI analysis, they spent around 80 hours validating findings manually, with static analysis across the ten stacks costing about $935 in total—roughly $2.58 per finding. Binary analysis, the report notes, was around six times more expensive and took twice as long as source-code analysis.
Daniel dos Santos, VP of Research at Forescout, said AI “dramatically compressed the work” and that AI-assisted workflows can uncover vulnerabilities in complex embedded software much faster and at a fraction of the cost, a shift with implications for critical infrastructure where legacy devices remain in service for years.
The article also stresses that AI is not a panacea. ReliaQuest’s research is cited as showing AI can enable cheaper, faster, and more scalable attacks, while some experts from Proofpoint have called AI-generated exploits highly flawed.
To mitigate risk, Forescout recommends security teams prioritise risk by reachability, exploitability, and operational impact; reduce exposure by disabling unnecessary services such as FTP, Telnet, IPv6 and web administration interfaces; isolate devices that cannot be patched; accelerate patching timelines where fixes exist; and ensure procurement includes supplier-defined support lifetimes, vulnerability disclosure policies, SBOMs, and ongoing security testing. The story is dated 16 November 2026.