Despite the fact that Infosecurity Europe was on, June’s #InfosecLunchHour brought together a varied group of cyber security professionals, OT and ICS specialists, GRC practitioners, risk advisers and security advocates for a conversation that moved quickly from the theoretical to the deeply practical. The topic was proposed and led by one of our regular lunch hour attendees, who stepped in to guide the discussion after I had to leave slightly earlier than usual.
Under Chatham House Rules, the session explored one of the more uncomfortable questions facing the sector right now: what happens when artificial intelligence stops being a tool that helps defenders and attackers in equal measure and starts actively steering opportunist attackers towards the targets that can do the most damage. The discussion ranged from operational technology and the Purdue model to Cyber Essentials, the cost of certification for small businesses and a live Ofgem consultation that everyone in the room was encouraged to respond to.
Setting the Scene: AI as an Accelerant
The session opened with the attendee leading the discussion setting out the concern that prompted them to propose the topic. They referenced a recent incident in Mexico in which AI was used to collate information from open sources and then direct attackers towards operational technology endpoints they might not otherwise have identified or prioritised. What troubled them, and several others in the room, was not that AI had broken anything new. It was that AI had lowered the barrier to finding and reaching the most damaging targets.
Their point was that a great deal of opportunist attack activity is relatively undirected. Attackers find a way in and exploit whatever happens to be in front of them. What AI changes is the ease with which an attacker of modest technical ability can be guided towards the systems where compromise has a physical, real-world consequence, without ever needing to develop that expertise themselves. The worry is less about elite, well-resourced adversaries, who have always been able to do this, and more about everyone else suddenly being handed a shortcut.
The OT Problem and a Definition Worth Agreeing On
Much of the conversation centred on operational technology, and it became clear early on that even agreeing what we mean by OT is part of the challenge. One participant noted that the term is used very broadly, despite significant differences in how security is actually applied across, for example, ICS and SCADA environments. The discussion lead offered a working definition they tend to default to, describing OT simply as technology that has a physical effect on the world, and then worrying about the more specific categories beneath that. As they put it, trying to list out ICS, SCADA, IoT, BMS, DCS, IACS and everything else every single time quickly becomes a headache.
The recurring theme was separation, or the lack of it. The point was made that many organisations still do not have proper separation between their OT and IT networks, which leaves them exposed in ways they often do not fully appreciate. The Purdue model, which is meant to provide that layered separation, is frequently referenced but, in practice, rarely applied properly. The gap between the architecture diagram and the operational reality is where a lot of the risk quietly sits.
Legacy Systems, Scale and the Shift from Discovery to Recovery
One participant shared research findings on the prevalence of legacy systems within OT networks and the very real difficulty of funding their replacement. These are not systems anyone would design today, but they are expensive, disruptive and sometimes unsafe to replace, so they persist. Another participant raised the challenge of securing OT infrastructure at genuine scale, using electricity transmission networks as an example of the kind of environment where the consequences of compromise are severe and the attack surface is vast.
A useful contribution came in the form of a diagram one participant shared, illustrating a shift in security thinking away from the proactive discovery of assets and towards a recovery focused strategy. The underlying point was that for smaller organisations in particular, the sheer complexity of securing OT can make exhaustive asset discovery unrealistic, and that planning properly for recovery may be a more honest and achievable starting position than pretending you can find and protect absolutely everything.
Third Party Risk and the Certification Question
The discussion moved naturally into third party risk, with one participant raising concerns about specialist suppliers and noting that the NCSC is working on measures in this space. Another participant asked a practical question that many organisations wrestle with what checks should actually be carried out when onboarding a third-party vendor. The response was that Cyber Essentials is currently the baseline expectation for government contractors, although Cyber Essentials Plus would be preferable given that it involves independent verification rather than self-assessment.
The discussion lead pointed the group towards the Cyber Resilience Pledge, which asks organisations to register with the NCSC, require Cyber Essentials certification and assign responsibility for cyber security at board level. There was also reference to the Cyber Security and Resilience Bill, with the expectation that organisations supplying into critical national infrastructure, data centres, defence and managed service providers should hold some form of cyber certification. One participant noted that at DCC level three, Cyber Essentials Plus is still mandated, and that is for the larger organisations rather than the small ones.
The Cost Barrier and the Idea of a Cyber Levy
This led into one of the more pointed exchanges of the session, on the cost of certification and who can realistically bear it. The discussion lead was direct about the barrier this represents for small businesses, particularly those without dedicated IT staff. They noted that Cyber Essentials Plus can run to around two thousand pounds a year for a micro business, a significant investment with no guarantee of winning the work it is meant to unlock, and that the scope for a Plus assessment can be as small as two laptops. Cyber Essentials itself was put at somewhere between roughly three hundred and sixty pounds and a thousand pounds. There was also a genuine concern that the assessor capacity simply does not exist to absorb any significant increase in demand.
One participant raised the figure of ninety million pounds attached to the cyber pledge, understood to be aimed at helping organisations achieve certification. Another floated the idea of a socio technical cyber levy as a way of funding security improvements across the wider ecosystem, rather than leaving the full cost with the individual businesses least able to absorb it. That participant shared related material with the group and would welcome feedback and challenge on the concept.
A Live Ofgem Consultation and a Call for Critical Voices
As the conversation turned towards supply chain regulation, the discussion lead flagged a consultation that Ofgem had opened on draft supply chain security guidance, released only the day before the session. Their ask of the group, and of the wider community, was specific. Rather than generic messages of support, the consultation needs genuine engagement and critical and constructive responses from people who understand the operational reality. Consultations of this kind shape what eventually becomes expected practice, and the quality of the guidance depends heavily on the quality of the responses it receives.
Cyber Essentials, 10 Steps and What Got Lost Along the Way
The final stretch of the discussion compared Cyber Essentials with the older 10 Steps to Cyber Security guidance and asked whether the sector had actually gained from the shift. One participant explained the distinction clearly: Cyber Essentials was designed as a technical standard built around five specific controls, whereas 10 Steps offered a broader, more general introduction to cyber security as a whole. They are different tools for different jobs, and treating one as a straight replacement for the other loses something important.
The discussion lead made the point that 10 Steps has not in fact gone away. It still exists and remains aimed at medium and large enterprises rather than SMEs, with the NCSC small organisations guide serving as the more appropriate starting point for smaller businesses. The frustration in the room was that none of this is marketed well. One participant described how most references now steer people away from 10 Steps and towards Cyber Essentials by default, which is both a shame and a missed opportunity. Another shared a recent experience of a charity that genuinely wanted to get security right but struggled to find anything aimed at them, until they were pointed towards the relevant NCSC guidance, which gave them most of what they needed.
There was some concern that IASME could end up creating too many specialised standards, adding complexity rather than clarity. On a more optimistic note, one participant felt that the partially achieved status within the newer framework could be genuinely useful, provided it is properly monitored and followed up rather than treated as a box that has been half ticked and then quietly forgotten.
Closing Reflections
As ever, the value of the session was in the honesty of the people in it. The throughline across the whole discussion was that AI is not so much introducing brand-new categories of risk as accelerating and directing the risks we already have, pointing them at the systems where the consequences are most physical and most serious. Whether the topic was OT separation, the cost of certification or the marketing of guidance that already exists, the same theme kept surfacing. The tools and the standards are often there, but the resourcing, the verification and the support to use them well are not.
Thank you to the attendee who proposed and led the topic with such depth and generosity, and to everyone who joined and contributed. A particular welcome to those joining the lunch hour for the first time. The willingness of this community to share real experience, including the difficult and unresolved parts, is exactly what makes it worth being part of.
A few things to take away from the session:
- The Ofgem consultation on supply chain security guidance is open now and would benefit from your considered and critical responses.
- The idea of a socio technical cyber levy is one that its proposer would welcome thoughts on.
- If you support SMEs, there is a clear appetite for accessible, well signposted help with Cyber Essentials adoption.
The next #InfosecLunchHour takes place on Wednesday 1 July 2026 at 12.30pm BST. If you would like to join a group of cyber and Infosec professionals for some relaxed chat over lunch, please contact me via lisa@unitysolutions.org.uk to be added to the calendar invite.
#InfosecLunchHour is a free, open, community networking event hosted by Lisa Ventura MBE FCIIS. All discussions take place under Chatham House Rules: participants may use information shared in the meeting but may not attribute it to named individuals or organisations.
