Listen to this episode:
About this episode
In this episode of Behind the Data, host Matthew Stibbe talks with Drew Heron — information security consultant at DigitalXRAID and a self-described governance and risk specialist. Drew explains why the hardest question he asks a client is simply “who owns this?”, why every organization’s data estate looks a lot like the drawer of shame in your kitchen, and why the companies that scare him most are the ones that ban AI outright.
The conversation covers kill switches and whether anyone would actually press one, the bank that lost a business-critical server in its own basement for 25 years, why most organizations tackle governance, risk and compliance in exactly the wrong order, and the one question you can only ever ask once: who’s going to prison if this goes wrong?
Meet the guest: Drew Heron
Drew Heron is an information security consultant at DigitalXRAID, now part of the Cypher group. He describes himself as a governance and risk specialist. His current work centers on AI governance, including ISO 42001 and the EU AI Act, alongside maturity assessments that map organizations against frameworks such as the NIST cybersecurity framework, ISO 27001 and Cyber Essentials. He has a particular interest in the human side of risk, drawing on behavioral science models including BJ Fogg’s B=MAT, and in reading the documentation, policies, and terms and conditions that most people skip, in order to understand how systems are actually connected and what data they quietly share.
Key takeaways
- “Who owns this?” is the hardest question in the room. There are usually three or four people with a stake and nobody with the decision. Ownership follows the budget: if you don’t control the spend, you can’t own the risk. You can delegate risk, but the delegate can’t own it.
- Your data estate is the drawer of shame in your kitchen. GDPR made most organizations glance in the drawer once. Cloud storage just bought them a bigger drawer. Now AI is rooting through it to see what’s in there, and “legitimate interest” is what’s keeping the 2018 takeout menus and the spare Allen keys in place.
- Compliance done first is doing 70 miles an hour in a blizzard. Governance is deciding where you’re going, who’s driving and who’s navigating. Risk is the weather and the state of the tires. Compliance is the speed limit. Plenty of breached organizations were fully compliant, because they started with the checklist and never established the destination.
- Banning AI outright is the most dangerous response. Staff will use it anyway. No technical control stops someone reading a screen and typing data into a personal device, so an amnesty and a culture change beat a block. Behavior can’t be changed directly, but motivation, ability and triggers can.
- CISOs are risk translators, not risk owners. A good security leader lays out the options and the consequences, the way a good lawyer briefs a client. The business still has to walk through the door and decide, and in most organizations that decision is deferred rather than made.
- This is no longer a technical problem with business impact. It’s a business challenge with technical considerations. That reframe changes who needs to be in the room and who signs off.
Chapters
- 00:25 Geeking out on documentation, policies and terms and conditions
- 01:16 The hardest question to ask a client: who owns this?
- 01:49 Kill switches, Jurassic Park and the lysine contingency
- 04:42 Governance and risk: ISO 42001, the EU AI Act and maturity assessments
- 05:48 The 7/7 commuter told to cycle instead, and other faulty risk management
- 07:54 Blast radius, and why a bad morning makes you a higher risk
- 09:09 The bank that lost a business-critical server in its own basement for 25 years
- 11:06 The drawer of shame: GDPR, legitimate interest and 2018 takeout menus
- 12:35 Hoarding 50,000 contacts nobody has emailed in a year
- 14:53 Why storage is about to get expensive again
- 16:24 The problem with a two-year-old DPIA
- 19:05 Legionella, budgets and why ownership follows the money
- 21:52 The rogue CRM nobody authorized, and the data that left HubSpot
- 24:35 Why you can’t fix rogue IT with technology, and what B=MAT offers instead
- 28:37 Doing 70 in a blizzard: governance, risk and compliance the wrong way round
- 31:16 CISOs are risk translators, not risk owners
- 32:20 “Who’s going to prison if this goes wrong?” — the question you can only ask once
- 34:39 A business challenge with technical considerations
Episode transcript
Matthew Stibbe (00:01) Hello, welcome to Behind the Data with CloverDX. I’m your host, Matthew Stibbe, and today I’m talking to Drew Heron. He’s an information security consultant from DigitalXRAID. Great to have you on the show, Drew.
Drew Heron (00:14) It’s great to be here, thank you for having me on, Matthew.
Matthew Stibbe (00:17) So I’d like to open this show, as regular listeners will know, with the question, what are you geeking out about at the moment?
Drew Heron (00:25) So it’s a strange one actually, because I go into how things are flowing through AI models and data models and things like that, rather than the actual models themselves, rather than actually building them. People sort of look at me kind of strangely — you like looking at documentation, you like looking at policies and terms and conditions? Yes, I am that weird person. But that to me enables me to understand how things work under the hood, how things are connected and what data they’re sharing with each other and things like that, to ask the questions that people don’t really want to talk about and don’t really want to answer. So that’s the kind of thing I’m geeking out about at the moment.
Matthew Stibbe (01:10) What’s the most challenging question for people to answer when you ask them?
Drew Heron (01:16) I think one of the actual hardest ones I encounter is normally who owns this? Because it’s normally about three or four people who say, we’ve got a stakeholder in this, or they’ve got an interest in this. But actually at the root of it, who has the decision? And that also leans onto sort of like kill switch and things like that, of what kill switch do you have on this AI system? Are you willing to press it if things go wrong?
Matthew Stibbe (01:44) If you’ve got your axe at the ready in case HAL goes rogue, right?
Drew Heron (01:49) Effectively, yeah. So it’s one of the things I like — one of my favorite movies is Jurassic Park, and they have one in there which is the lysine contingency. And when it’s brought up they say, no, we can’t push that at all. But that’s your kill switch. That’s the button you press that, yes, you destroy the park, but effectively you save a lot of lives.
Matthew Stibbe (02:17) So jumping deeply into this, this question of kill switch is really important, I think, because you kind of also have to engineer the decision logic for when you’re going to press or pull it, I guess. That’s the bit that’s hard, isn’t it?
Drew Heron (02:37) Yeah, 100%. So even going back before AI, even on the normal systems and systems we’ve had for 30, 40 years, when do you pull that internet plug?
So I’ve worked with organizations where they’ve done it quite quickly. I can remember doing a tabletop with a medical facility and they were really quick. It’s like, yeah, this is going a bit pear-shaped. We’re just going to pull the plug to the internet. And that was even before we got to the crisis point of the tabletop, which was really encouraging to see from them. But like you said, unless you’re working through this, it’s quite an interesting question to ask people as well when you’re asking about questions of when do you press it, who’s got the authority to press it, and are they going to be overridden by any of the incentives and things that come through normal business.
Matthew Stibbe (03:31) Easy to be brave in a tabletop exercise, much harder when it’s real life, I think.
Drew Heron (03:36) Indeed, indeed. When you’ve got sometimes millions of pounds in an investment of product and you’re there to pull that switch to kill that product, causing reputational damage, financial damage and even personal career damage, are you willing to push that button?
Matthew Stibbe (03:56) Yeah. I’m a big space enthusiast and I always have in my mind, when you see the pictures, the film of the control room, when they’re launching a rocket, there’s somebody sat there, typically in the front row, who’s basically got a button that they press if the rocket is going to fly into Orlando and kill people. They press it and the rocket blows up. And that’s a hell of a responsibility because it’s multi-millions of pounds of rocket. But if there’s people on it… And they have to have really clear rules for it. I’m sorry, we’ve gone down a rabbit hole of kill switches. Let’s bring this back to you. Tell me, what do you do at DigitalXRAID? What’s your sort of area of expertise?
Drew Heron (04:42) So yeah, as you’ve probably guessed, I’m doing a lot of work in AI at the moment, sort of increasingly working around the ISO 42001 and the EU AI Act that’s recently come in. It’s just getting stronger and stronger. But before that, concurrent with that, I’m doing a lot of maturity assessments for organizations, mapping against things like the NIST cybersecurity framework, NCSC, ISO 27001, Cyber Essentials, all that good security work.
But my actual specialty, I always say, is I’m a governance and risk specialist. So governance of data, governance of how things are done, why we’re doing things. And that’s the most important thing to me, is why are we doing things. And then the risk management of that as well. I can talk for days on the risk management side of things, but I won’t bore you on that. But I’ve got a whole utopian view on how this should be done, and the baby steps to get there.
Matthew Stibbe (05:48) I’ve always had a sort of curiosity about risk management and an amateur enthusiasm for it. But I remember a friend of mine after the 7/7 bombings in London, he used to go on the underground to work every day from Notting Hill, and his wife said, right, no more underground. You’re going to cycle to work every day. And I just thought, you know, that is just faulty risk management right there, right? Well, just in case you didn’t know, cycling in central London’s an insanely dangerous thing to do. You might as well do off-piste skiing or something.
Drew Heron (06:19) Yeah, I used to be a road cyclist and I just won’t ride on the road anymore because I just don’t feel safe. So I’ve gone onto mountain biking and gravel instead. That’s my route. To me, that’s just a whole termination of risk. So there are four things: you either accept it, you terminate it, you transfer it or you tolerate it. That was just a whole termination of risk. But again, risk management is just so fundamental to, I think, daily life. Whether you work in data, IT, security or anything, people don’t realize just how much risk management they do day to day. Even crossing the road, you’re doing that risk management in your mind. Is it safe, is it not? What’s the likelihood of a car coming down and hitting me? What would be the impact of that?
I always think there are two in information security or IT that are missed on that. So I think it should be a fuller thing. I think we’re going off topic. But I always think it’s like you’ve got the likelihood of something happening, the impact across the entire business, but also, especially in the modern era, you need to kind of think of, how quickly is that going to affect my business?
Drew Heron (07:38) Is it going to be under two hours or is it going to be four years? Because they’re two very fundamentally different risks, even though the impact to the business might be the same. And then, yeah, how fast you can recover and then remediate, yeah.
Matthew Stibbe (07:47) Yes. Yeah, so certainly affects how quickly you could remediate it.
Drew Heron (07:54) And then you also have to look at the blast radius as well. How wide is this going to affect my company? So is it going to affect everything or is it going to affect just one device or one person or things? But even then we could go into the behavioral science of that as well, which brings in the sort of human risk management of things, because people who have had a bad day… So you wake up in the morning, your kids don’t want to go to school. So you go to get them ready and you’ve got somebody cut you up on the drive or just rude to you, spill your coffee, you’re going to be in a bad mood, you’re more likely to make mistakes, so therefore you’re at a higher risk of doing something like opening a phishing email or sharing your password with someone, than if you’ve had the best start to your morning you could ever have.
Matthew Stibbe (08:46) I mean, everything you’re talking about, anyone who’s involved in data engineering probably needs to be thinking about this stuff and baking it into their thought processes. But I’m interested: when you go into an organization with this mindset that you have, what’s the first question you ask about their data and what’s the answer you usually get?
Drew Heron (09:09) So this kind of comes back to when we’ve spoken sort of earlier a few times about like GDPR and things like that. And one of the first questions I always ask is, do you actually know where your data is? Because a lot of people understand where their servers are. Interesting stories I have about that. A company, a fairly well known bank, that didn’t know they had a server somewhere, or the physical location of the server. They knew it existed; they just didn’t know where it was. Indeed, they actually found it in the basement somewhere, and it hadn’t been touched for 25 years, but it was vital to their organization. We’ll put it that way. But they had no idea where it was. It was quite interesting when you look at it like that.
A lot of organizations don’t know where their information is in the cloud. But what server, who has access to this data? Why do we have access to this data? Do we actually need this data? Is it still required to do what we wanted it to do? Or even if it’s even real data anymore.
So one of the best sort of analogies I have for this is the rubbish drawer you have in your kitchen. So everybody has a rubbish drawer and I will die on a small hill. I think, especially men, we all have the drawer of shame, the bag of cables or box of cables that you just, I may use it at some point. And you might have something like that, and everybody, I’m sure, has got it in their head now if they can see it. Might be over on your desk somewhere.
Matthew Stibbe (10:46) The drawer of shame.
Drew Heron (11:06) When GDPR came in, we sort of looked through that drawer a little bit and that was about it. You knew where that drawer was, you kind of looked at what was in it and thought, okay, well, so that’s everything. The worst thing that ever came out of GDPR was legitimate interest, which is just such a gray area.
Matthew Stibbe (11:25) The drawer is actually labeled legitimate interest, isn’t it, for PII?
Drew Heron (11:30) Yeah, pretty much. Yeah, for PII and things like that. It’s legitimate interest. We don’t know when we’re going to use it, but we have a legitimate interest to retain all this data. So in that drawer, you’ve got sort of like 40, 50 takeout menus from 2018 or 2019, takeaways and things like that. And you’ve got bits and bobs of like nuts, bolts, Allen keys from an IKEA order you might have bought 10 years ago, but nobody’s actually looked into it. And all that we’ve done now, especially in the GDPR and data centers and things like that, and with the ability to scale in cloud, is just get a bigger drawer. So the drawer is no longer your sort of standard, I think it’s a 50 mil, 500 mil sort of drawer. It’s now like a meter long drawer, or a meter wide.
And all that we were talking about with AI and things like that is AI is now opening that drawer and rooting through that to see what’s in there.
Matthew Stibbe (12:35) I see this a lot. I mean, in my little world I go into companies and look up their HubSpot CRM portals and things, and you might have 50,000 contacts in there, and you know, people are paying HubSpot to store these contacts. I mean, it’s not cheap. And then, you know, 20,000 have never been emailed; 10,000 haven’t been emailed for a year. Of the rest, a third of them are actually undeliverable; they’ve bounced, or they’ve opted out. But nobody wants to throw away the data. It’s sort of like a miser’s hoard. I don’t know if you’ve seen these TV shows where you go in and there are people have hoarded all this stuff in their house and they don’t want to clean it out. How do you… I mean, from your perspective, all of that’s an attack surface; all of that’s a regulatory compliance risk. How do you go about sorting that out? What’s the strategy for, I don’t know, Marie Kondo-ing that stuff?
Drew Heron (13:27) See, that’s the difficult thing, because it’s almost you trying to pry data out of people’s really tight-fisted grip on it. And we all know the departments that we’re talking about here, because it might be useful in five years. It might just be, it’s that cable again. With how, say, recruitment is at the moment and how tenures are in organizations, if somebody hasn’t looked at their old contacts in the CRM, in about six months to a year they’re probably not working at that job anymore, or they’ve moved up or moved on and things like that. And it was one of the conversations I was having with our own marketing department at one point, and it was a case of, how much do you put into that? How much do you really need that data from three years ago to kind of reach out and to continually bombard with an outreach email or information on that? That’s, like you say, just opens you up to that risk of repercussion at the end of the day.
Matthew Stibbe (14:44) And the issue is, of course, storage is cheap, and attention and discipline and diligence are expensive, because it’s time. It’s people's time to do it.
Drew Heron (14:53) I was going to say, although storage is increasing in cost at the moment, because of the increased reliance on AI. So storage is going to become very, very expensive. So I can see a lot of cost savings coming from people effectively doing like a spring clean of their data.
Matthew Stibbe (15:19) In order to have less context, less token consumption as you feed that into AI. Is that what you mean?
Drew Heron (15:25) Exactly. Yeah, and then even the actual overall storage. If you’ve got a thousand terabytes, that’s going to be a lot of storage costs compared to, say, 500 terabytes in a few years. And we’re seeing this in the consumer market, where if you look at even building a new PC, RAM and storage is extortionate. So I can see that transferring into the corporate world not too long from now.
Matthew Stibbe (16:02) You expressed a fairly controversial view, and you tiptoed up to it just now, about GDPR. What’s the risk or the danger behind a sort of superficial compliance but deeper non-compliance with it? And I’m not talking about the legal risk, but just…
Drew Heron (16:24) Yeah, I’m not a legal expert so I don’t want to start giving legal advice on things, but effectively you are running the risk, if you’re not doing those DPIAs, if you’re not doing that actual due diligence on things, you are running the risk that when things happen, you’re up against the ICO. And the ICO don’t really like… but they do come down quite hard when they need to. So yeah, the whole thing of, we did a DPIA, means precisely nothing, and it’s not what the data is doing now. So even historically, yeah, we’ve done a DPIA, or a data protection impact assessment for those who aren’t familiar with the acronym. If that is 12 months, two years old, it doesn’t reflect what happens now. So you need to be reviewing that constantly, and even the evolution of…
Start with the business impact assessment. Do we actually need this system, this data and things like that? Then if you’ve got personal information in there, that’s when you go for a data protection impact assessment. And then if you fit an AI onto that, you need to kind of do an AI impact assessment as well. And how far does this spread? One of the things that I think people overlook with not even AI, but systems as a whole, is the API connections, where they just grab data. So they may grab an entire database where they only need three columns. So by grabbing the entire database, you could run the risk of getting more data than you require, which technically could be a GDPR breach.
Matthew Stibbe (18:18) I audit a lot, or review a lot of HubSpot portals for many people, and it’s often a huge surprise to the people I talk to how many different people in the organization have hooked up a ChatGPT or a Claude MCP connection into their data using the permissions they have as a user. And you know, it’s like, okay, how do we feel about that? So how would you, if you step back from that from a sort of governance or planning perspective, how would you recommend an organization should sort of put in place the governance on top of that, the kind of the principles or the ownership of it?
Drew Heron (19:05) In terms of looking at it, so to me, I always think ownership of pretty much most things comes down to budget owner. So if you don’t control the budget, you can’t own the risk. The money comes down to risk. So you can delegate that risk, but you can’t own it. The delegate can’t own it. But you need to have that ability to change the actual underlying controls to then own it.
So for example, there was actually an interesting case years and years ago. I think this was back in the 80s, I think it was at Barrow-in-Furness. There was a leisure center and a facilities manager. And they brought in a facilities director who decided that, to save costs, we would not service the air conditioning as regularly. So you can already see where this is going, with stale water. And yeah, so turns out there was a Legionella outbreak, and I think it was four or five people may have lost their lives on this one. So yeah, when it came to court and such, the facilities manager of the leisure center went up in court, they talked to them, so asked were you in charge, and went through a whole court system. But he didn’t control the budget. The facilities director controlled that budget, and it was ultimately their responsibility and their accountability to determine how often that budget, or how that budget, was spent. So it was deemed that they were actually fully responsible for that risk. So that’s where, to me, the whole, like we were saying, the ownership piece of this. And then come back to data. So the owner of the data comes back to: if you can control that data, then you own it. So if you choose where that’s kept, how that’s kept, who has access to it, you’re the owner. And it kind of goes into the data protection side of things, well, that’s when you are the custodian, not the processor and things like that. But ultimately again, you’re responsible for that data.
Matthew Stibbe (21:33) I think this- who owns things and how things kind of find their way into an organization- is interesting. You told me a story about an application that you found on a client’s Microsoft tenant that nobody knew about. Can you tell the story for the people?
Drew Heron (21:52) Yeah, so we joined an organization. We were doing a, they call it a cloud configuration review of their tenant. We were going into the Microsoft application stack and looking at what applications they had connected to their tenant. And we found one, which was, I can’t remember the exact name, but it was a CRM of an organization. So we did some more digging and we found out the CRM was an American company. I’m not sure if this is very strange; the company itself used HubSpot. So why have we got another CRM on the system that isn’t HubSpot? Why would we have that? So we did more digging and a bit more investigation. It turns out that one of the people in a department — I’ll let the listeners infer which department that was.
Matthew Stibbe (22:53) Marketing.
Drew Heron (22:59) No, it wasn’t actually. It was marketing adjacent.
Matthew Stibbe (23:02) Marketing adjacent. Right. For once we’re not the responsible party. All right, good.
Drew Heron (23:15) They’d been signed up to a webinar and got free access to a CRM. And what we found out through looking at their logs and things, they actually had signed up to this free trial and they were actually downloading client information from HubSpot and uploading it to this other CRM to use that functionality of that CRM.
Yeah, that was an interesting find. Lots of quick conversations. I think the first question we asked was, who authorized this? Who allowed you to go and do this?
Matthew Stibbe (23:52) Probably the answer was nobody, I just did it.
Drew Heron (23:55) Effectively, yeah. The other comment was, I didn’t think it was a problem; I was performing my duties, I was doing my job. Because at the end of the day, people just want to do their job and do it well.
Matthew Stibbe (24:10) You can put it into an IT policy: don’t install software or put personal information or PII or whatever into systems without, you know, it’s authorized, and you can say that all day, people still will do it. So what’s the technical answer to avoiding that problem? How do you avoid rogue IT, rogue data pools, I suppose is a way of thinking about it?
Drew Heron (24:35) I think this could be the answer that nobody wants to hear, but you can’t. There’s no real technical way of doing it, because there’s always a way you can… You could stop them…. You could put something like a cloud access security broker in front of it to stop them uploading things for that. But then you’ve got to look, okay, so how do you get an email out? So they could email it out and then upload it to a different CRM. So you block the email to personal devices. But then, depending on your business- because I’m sure, Matthew, you have to email personal accounts at some point. So it stops you doing that. So you can’t put that technical thing in there. Also, you can’t really stop somebody with a personal device next to them just typing it in while looking at a screen. And that comes back to culture. So you need the culture and the right thing to do, effectively.
There’s a really interesting sort of behavioral science formula that I always like to think of, and it’s done by a guy called BJ Fogg. And what he put down is B equals MAT, or M-A-T, where you can’t change the behavior directly. And I can agree that you can’t really affect people’s behavior, but you can affect MAT. And what MAT stands for is motivation, ability and trigger. So what motivates somebody to do something? You can change that, you can affect that. The ability is not so much can they do it, but also how easy is it to do something, what enables them to do things. And then the trigger, what makes them do it. And I think that that’s really missed sometimes. That everybody tries to change the behavior, but not actually anything about the motivations, the abilities and triggers.
And this is where I think a lot of the old analogy of water, the water goes through the easiest cracks and will always find a way through things. If you think of people’s work like that, they just want to get the job done. A lot of people just want to turn up to work, get their job done, go home. They’re not looking at, well, if I do this it’s going to be risky for the company or less risky. It doesn’t affect them, they just want to get the job done and not get fired. At the end of the day you want to keep your job, so you’ll do whatever you can to make your job easier. I think this is where the conversation with AI and data comes in, because we have the amnesties, as I always say to people when we start talking AIs and things, of tell us what systems you want to use or are using, and then we’ll see if they’re suitable. Because I’ve said a few times to people that the companies that scare me the most, especially today with AI and things, are the people who block it outright, because your staff or your employees are using it regardless.
Matthew Stibbe (27:59) Yeah. It’s like the old thing of, in the old days you used to enforce very strong passwords and change them every thirty days or ninety days or something. So what happened? People wrote their strong passwords on a post-it and put them on the monitor. You know, you’ve ended up making…
Drew Heron (28:11) Or enumerated them. Enumerated them. Instead of having like… Yeah.
Matthew Stibbe (28:16) Yeah, like strong password one, two, three, four. Yeah. So yeah, you end up with an unwanted outcome trying to solve it by brute force. This sort of brings me on to a point that you made earlier with me, which I thought was very interesting. Taking governance, risk and compliance, you suggested that people do it the wrong way round.
Drew Heron (28:37) Yeah, so I’ve seen a lot in the organizations I’ve worked with and just around the world and the industry at the moment, because you see companies getting breached daily, but they are all ISO 27001, Cyber Essentials and things like that compliant. And as we were talking earlier, compliance is a checkbox. It’s a checklist of things you need to do, whether you actually need it for your business or not is a different story.
So as we were saying, my favorite analogy for this is the car journey and things. So the governance of it is: where are you going and how are you going to get there, who’s going to be the navigator, who’s going to drive the car, who’s going to fill the car up, all that good stuff. And then the risk management is, well, what’s the weather going to be like, what’s the road surface going to be like, do we have old tires, is the car going to blow up if we drive it down the road five meters? And then the compliance is speed limit, MOT. And I find a lot of organizations, if you flip it, they’re all the way around. It’s, we’ve already decided, right, we’re going 70 miles an hour down this road, it’s absolutely in a blizzard, and they have no idea where they’re going or who’s even driving.
Matthew Stibbe (30:01) So they’re compliant but they’re at risk.
Drew Heron (30:06) Yeah, fully at risk, because what they’ve done is they’ve looked at the checklist, said we need an acceptable use policy. Okay, so they’ll go on the internet or ChatGPT, Claude, and get a templated acceptable use policy, get it all ratified, authorized, put it out there. But it might have something like, well, the acceptable use of, I don’t know, like HubSpot, but your company uses Salesforce. So it doesn’t marry up. It stops there. And then on to sort of risk management. We were talking earlier about our whole risk management and sort of decisions. I came to the comment of, I think these decisions are just deferred. They’re never really made in an organization, and especially around risk. Nobody likes to take a risk decision.
Matthew Stibbe (31:09) Yeah, there’s a danger that you end up calling somebody’s baby ugly and their mother stupid, right?
Drew Heron (31:16) Yeah. And it’s always been seen as an IT problem as well, and I’m sure the people, our audience, will fully understand that as well with the data. IT has always been almost like the hidden child of, well, that’s who risk sits with. It’s all IT problems. And information security has grown, data management and… Effectively, InfoSec is just a subsection of data, which is a subsection of IT.
Matthew Stibbe (31:51) So one last question I think. If you are in IT doing data engineering and this lands on you, what can you do to, one, think at a higher level of abstraction about GRC? And two, what can you do to educate the rest of the organization that it isn’t only just a technical fix and I’ve got some piece of software that’s going to solve it for you?
Drew Heron (32:20) Yeah, it’s difficult because, again, a lot of organizations see it as an IT problem. But the thing I would always start with is governance as the most important starting point, and from that is why, the question of why. Why are we doing this? Why do we need this data? Why do we need this connection between these two points?
Then you can start looking towards the, well, who owns this, who’s accountable, who is going to be the one. And one of my favorite questions when you start moving into that risk decision of ownership and accountability, and you can only ever use this once, is who’s going to prison if this goes wrong?
That’s the one question you can have in your back pocket. And like I said, you can only use it once, because if you keep on using it, it just loses all kind of response like you’ve just had there. Yeah. But that’s the kind of thing I would start with.
Matthew Stibbe (33:19) Yeah, it’s crying wolf the second time. There’s a sort of subversion of that, which is who gets fired if this goes wrong. So slightly less…
Drew Heron (33:35) Yeah, it’s less on the nose I think, but I think they have the same response of, well, they would actually be on the chopping block if this went wrong. So it’s like, well, okay, then they own the risk. I find this a lot with a lot of security leaders, CISOs and such, as their risk translators or risk communicators. They’re not risk owners. So they can inform you what all the risks are, can tell you how bad this is going to go if it goes wrong, but not make the decision to kind of go, okay, these are the ones we’re going to implement, this is the way we’re going to go down. They can give you a choice of A, B, and there’s always a plan D, all the way down to Z, and then you know the numbers if you really want to. But from there, they can give you the options. You have to walk through that door.
Matthew Stibbe (34:35) Yes. Yeah, like a good lawyer advising a client, right?
Drew Heron (34:39) Exactly. I think one of the things I got opened up to from someone I was talking to a while ago was, information security and data and IT is no longer a technical problem with business impact, but it’s a business challenge with technical considerations.
Matthew Stibbe (35:05) That’s a very good place I think to bring this conversation to an end, which I’ve enjoyed enormously. Drew, how will people find you and DigitalXRAID online?
Drew Heron (35:16) So they can find DigitalXRAID at digitalxraid.com. We have a parent company. We are now part of a bigger group, Cypher, cypher.co.uk, as well. And I’m also on LinkedIn as Drew Heron, to be honest as well. So if people do want to reach out, have a conversation, by all means, do it through either one of those three channels.
Matthew Stibbe (35:38) Fantastic. Well, Drew, it’s been a delight having you on the show. Thank you very much.
Drew Heron (35:42) It’s been absolutely fantastic. If you want me to come back and chew your ear off, I’m happy to do so.
Matthew Stibbe (35:49) And on that bombshell, happy to have you back. But that brings this episode to an end. If you are listening and you’d like to know more about CloverDX or listen to more podcasts or get practical data insights, we’re online at cloverdx.com and the podcast is cloverdx.com/behind-the-data, and available from all good podcatchers. Thank you very much for listening and goodbye.
Resources
- DigitalXRAID — Drew’s employer, a UK cybersecurity firm
- Xypher — DigitalXRAID’s parent group
- Drew Heron on LinkedIn
- ISO/IEC 42001 — the AI management system standard Drew’s current work centers on
- EU AI Act
- ICO guidance on data protection impact assessments — the DPIAs Drew argues most organizations treat as a one-off
- Cyber Essentials — one of the frameworks Drew maps organizations against
- BJ Fogg’s behavior model — the B=MAT formula Drew uses to explain why blocking tools doesn’t work
Related content
- Data quality with CloverDX
- Blog: How to establish a strong data governance framework
- Webinar: How to effectively migrate data from legacy systems
More from Behind the Data
- 80% of your data is dead: lessons from 35 years behind the data - with Peter Aiken PhD
- The human impact of data transformation - with Todd Schnirel and Jennifer McCalicher of Airline Hydraulics
