TWO enterprise cybersecurity leaders trialled AI in their SOCs for six months, sharing their findings at RSAC 2026 Conference. Mittal, overseeing a Fortune 500 food manufacturing company, reported that mean time to discovery improved 26% to 36%, mean time to response dropped 22%, and false positives fell by 16 points, with guardrails such as enforced citations, human approval gates, tool allow lists and full audit logging.
Gupta, at a financial organisation, found AI sped up fraud detection, underwriting, algorithmic trading and risk modelling, but a two-week test on a non-production system where AI was given full control produced negative results, including an instance where AI removed users from the system. Mittal emphasised that AI was embedded as a read-only triage assistant that synthesises alerts from multiple sources and would not directly interact with PLCs, SCADA or production equipment.
The pilots also highlighted ongoing challenges, such as additional false positive alerts and the need to manage AI across sprawling OT and legacy systems, with human final action still required. 23 March 2026.