25 Years of Warning Signals: From OT to AI
About This Session
Twenty-five years ago, operators were already asking what happens when the attack aims at water, power, and trains. The industry wrote frameworks. The adversary wrote plans. Warning is not watch out. It is a discipline, and it is how we get left of boom before machine-speed attacks turn detection into an obituary.
This keynote is about the discipline that closes that gap. It was born in the intelligence community, where analysts spent decades on one question: why does surprise keep happening when the indicators were already in hand? The answer applies to every plant floor in this industry. Warning is not watch out. It is a judgment call about probabilities, made before the evidence is conclusive, by a person willing to be wrong on the record. Operators have practiced it their entire careers. It has been called experience. It deserves its real name.
The talk traces the discipline from the steam plant to the intelligence community to the plant floor, walks it through the adversary campaigns of the last five years, and lands on the question this decade decides: whether agentic AI, with human judgment architected in, becomes the first genuine chance to get left of boom at machine speed. The alternative is detection at machine speed, and that ends the way it has always ended.
This keynote is about the discipline that closes that gap. It was born in the intelligence community, where analysts spent decades on one question: why does surprise keep happening when the indicators were already in hand? The answer applies to every plant floor in this industry. Warning is not watch out. It is a judgment call about probabilities, made before the evidence is conclusive, by a person willing to be wrong on the record. Operators have practiced it their entire careers. It has been called experience. It deserves its real name.
The talk traces the discipline from the steam plant to the intelligence community to the plant floor, walks it through the adversary campaigns of the last five years, and lands on the question this decade decides: whether agentic AI, with human judgment architected in, becomes the first genuine chance to get left of boom at machine speed. The alternative is detection at machine speed, and that ends the way it has always ended.
Speaker
Michelle Farr
VP, CISO - NXP Semiconductors
Michelle Farr is Global Chief Information Security Officer (CISO) and Vice President at NXP Semiconductors, where she runs security as a warning discipline, carrying the analytic tradecraft of the intelligence world into commercial practice so judgment arrives ahead of the incident.
A Navy Mustang who earned her commission from the ranks and was among the first women to qualify as a Surface Warfare Officer while operating in an expeditionary warfare environment. She went on to serve across the Intelligence Community, including as Senior Advisor for human intelligence at ODNI. She chairs the Government Security Committee at XTAR and is completing doctoral research at Purdue on how institutions calibrate trust in algorithmic judgment. She has spent her career on the boundary where commercial enterprise meets national security obligation, and she has never worked it as a spectator.
A Navy Mustang who earned her commission from the ranks and was among the first women to qualify as a Surface Warfare Officer while operating in an expeditionary warfare environment. She went on to serve across the Intelligence Community, including as Senior Advisor for human intelligence at ODNI. She chairs the Government Security Committee at XTAR and is completing doctoral research at Purdue on how institutions calibrate trust in algorithmic judgment. She has spent her career on the boundary where commercial enterprise meets national security obligation, and she has never worked it as a spectator.