Forescout AI Vulnerability Discovery
3 steps1 pieces3 Sources
Forescout Technologies, a US-based cybersecurity firm, announced on November 12, 2026, that its research division, Vedere Labs, used off-the-shelf AI models to discover 66 new vulnerabilities in embedded TCP/IP protocol stacks within 90 hours. This represents a significant reduction from the approximately nine months required for similar manual analysis during the company's earlier Project Memoria (2020-2021). The study analyzed 10 popular TCP/IP stacks, identifying 99 total vulnerabilities, of which 52% were rated high or critical severity. The automated static analysis cost approximately $935, averaging $2.58 per finding, while dynamic validation took about 80 additional hours. Fifty-four of the new vulnerabilities were reported to vendors and maintainers; as of the report's release, 19 open-source issues had been patched, but only five proprietary software fixes were completed. Forescout emphasizes that while AI dramatically compresses discovery time and cost, it does not eliminate the need for expert researchers. The findings highlight a shifting economic landscape where both defenders and attackers can leverage AI to find flaws in foundational software components used across OT, IoT, medical devices, and critical infrastructure. Supply chain risks remain complex, as patches must propagate through multiple vendor layers, and legacy devices may never receive updates.
How it got here
November 12, 2026: Forescout Technologies releases press announcement detailing how its Vedere Labs team used AI to find 66 new vulnerabilities in 90 hours, reducing analysis time from months to days at a fraction of the cost.
forescout.com ↗November 13, 2026: Industrial Cyber reports on the findings, noting that 52% of the new vulnerabilities are high or critical severity and highlighting the implications for operational technology and IoT security.