Cyber security | 16 November 2026

Industrial Security Conference Copenhagen: Key Takeaways

Published by Ashleigh Buchan

The Industrial Security Conference in Copenhagen brought together speakers from across industrial cyber security, covering network visibility, engineering resilience and the growing role of AI in operational decision-making. I attended the event last week, and a few discussions stood out for their relevance to industrial operators.

The Industrial Security Conference in Copenhagen brought together specialists from across industrial cyber security to discuss the challenges facing critical infrastructure operators, from network visibility and engineering resilience to the growing influence of artificial intelligence.

I attended the conference last week, and three presentations in particular caught my attention. Here's what stood out and why I think they're relevant to industrial operators.

Improving OT visibility without disrupting operations

Roxana Magdo of Cubro Network Visibility explored the practical challenges of monitoring industrial networks, where ageing infrastructure and operational constraints make changes difficult.

Visibility matters in OT cyber security, but introducing additional monitoring tools into a live industrial environment isn't always straightforward. Changes to network configurations may require engineering approval, testing or planned maintenance windows, especially where production availability is critical.

A case study involving a European electricity operator illustrated this challenge. The organisation had legacy infrastructure, remote substations and multiple security and monitoring tools requiring access to network traffic. Adding further monitoring capabilities risked introducing additional complexity into the production network. The approach presented used passive network TAPs and packet brokers to provide monitoring tools with the traffic they needed, reducing the need for repeated changes to production network configurations.

The example showed how operators can improve network visibility while minimising changes to live production systems.

Engineering for disruption and recovery

Patrick C. Miller, President and CEO of Ampyx Cyber, approached industrial cyber security from a different perspective in his presentation, Everyone Has a Plan Until They Are Punched in the Face. His discussion of Cyber-Informed Engineering focused on designing systems to withstand disruption, contain failures and recover safely when preventive security measures are no longer enough.

One point that stood out was the importance of maintaining independently verified configuration baselines and recovery procedures that do not depend entirely on infrastructure which may itself be unavailable or compromised. Patrick discussed keeping trusted configuration records and rehearsing recovery scenarios, including situations where normal services such as Active Directory or DNS cannot be relied upon. There's also a practical consideration for industrial operators. Having backups is necessary, but organisations also need confidence that they hold the correct, approved versions and can access them when required.

For sites operating complex or ageing control systems, recovery may depend on specialist engineering knowledge, legacy software and configurations that have changed over many years. If the information needed to rebuild a system is scattered across different locations or dependent on particular individuals, recovery can become significantly more difficult.

His presentation raised an important question: could an organisation still recover its control systems if the infrastructure it normally relies on was unavailable?

AI and the responsibility for operational decisions

Earl W. Shockley, President and CEO of INPOWERD LLC, explored how artificial intelligence is changing decision-making across the energy sector. He discussed the opportunities for AI to strengthen human decision-making across energy operations and planning, alongside the need to understand its limitations and maintain appropriate oversight.

One of the concerns he raised was the risk of false precision. As AI systems produce increasingly sophisticated recommendations, there is a danger that people begin to accept those outputs without sufficiently questioning the information or assumptions behind them. In critical infrastructure, where decisions can affect reliability and public safety, that presents a significant challenge. A recommendation may appear convincing, but operators still need to understand how it was reached and whether it is appropriate for the operational circumstances.

Earl also examined where AI should sit within decision-making processes. While activities such as monitoring, routine analysis and scenario generation may be delegated, he argued that responsibilities involving public safety, reliability risk acceptance and regulatory accountability must remain with people.

As AI becomes more widely adopted across industrial operations, organisations will need clear boundaries around its use, including who validates recommendations and who is ultimately accountable for decisions. My own view is that this makes the quality of operational data matter more, because it shapes the recommendations AI systems produce.

What can industrial operators take away?

Roxana Magdo demonstrated the practical challenges of obtaining better network visibility without affecting production systems. Patrick Miller examined the importance of trusted configurations and recovery arrangements when disruption occurs. Earl Shockley considered the risks of allowing increasingly sophisticated technology to influence operational decisions without sufficient human oversight. For industrial operators, these are familiar challenges in many respects. Engineering teams already work within constraints around system availability, change approval, ageing equipment and the need to maintain reliable operational information.

If you couldn’t attend, these are three questions worth taking back to your own site:

  1. Could we restore our control systems if Active Directory, DNS or our normal network services were unavailable?
  2. Do we hold independently verified configuration baselines, and do we know where the approved versions are and who can reach them?
  3. For any AI-assisted recommendation, do we know who validates it and who is accountable for the decision?

Coming to these sessions from the marketing side of an industrial software company rather than as an engineer, what struck me was how often the practical answer came back to information. Network visibility, recovery planning and AI-assisted decisions are quite different subjects, but each depends on being able to trust what you know about your systems and reach it when normal services may not be available. That is the area Asset Guardian works in, through software and configuration records, change history and controlled backups, and it was good to hear speakers from other disciplines describe the same need from their own direction.

Ready to get started? Discover Asset Guardian