Adopting AI to Protect Industrial Control Systems: Assessing Challenges and...
summary
Paper reports 18 interviews with ICS practitioners (13 operators, 5 vendors) and cites a survey where 10% using AI and 19% experimenting. It documents highly heterogeneous alarm deployments (PLC, sensor, SCADA), regulatory limits like NERC, and that human-heavy tasks—especially alarm diagnosis and alarm management—consume months of effort. Recommends deploying AI as assistive tools for diagnosis/management, new decentralized hosting models, and interactive pilots to show transparency and build trust.
AI-written summary · about
chapters (8)
- 0:00AI for industrial control systems
- 1:13Securing industrial control systems
- 3:20AI research versus adoption
- 5:26Study design and interviews
- 6:29AI adoption challenges
- 8:33Human factors in alarms
- 10:41Practitioners on AI in ICS
- 12:25Summary and recommendations
related talks
01h 03mTrends and Top Takeaways
46mAI in Cyber: Is the Cyber Profession Ready for Its Impact?
42mAI for security - friend or foe?
15mAI Security Simplified: Turning Complexity into Cohesion
30mGenAI Opportunities and Challenges: Where 370 Enterprises Are Focusing Now
52mAI Governance: The Security Perspective
07mCybersecurity Analysts' Perception of AI Security Tools and Practical Implications
36mModus OperandAI: Practical Security for Artificial Intelligence