Dell’s Rob Emsley on AI in Data Protection

hi I'm James Maguire and on today's webcast we're taking a look at data protection and artificial intelligence in this new era of AI data is more important than ever really data is the very Foundation of AI but the importance of data raises all sorts of questions about cyber security how do you protect your data in this new era to answer that question I'm joined by Rob emsley director of data protection product marketing at Dell Rob very good to have you with us today yep lovely to be with you James always good to talk about my favorite topic of data protection great so I think it's interesting about AI I think sometimes uh companies think oh you know we can bring in some AI sprinkle over the software it's all going to be a good but it but it turns out that hey data is is super important when it comes to ai ai goes nowhere without data which means of course protecting data becomes more important than ever how do you see the evolving cyber threat landscape here in the age of AI yeah well I think that um we're living in interesting times with respect to um AI being a an aid to uh both organizations and vendors that develop data protection solutions to uh to help uh deliver um cyber security and more importantly cyber resilience in a better way but also um the emergence of generative AI uh is something which unfortunately the Bad actors uh can also leverage you know I think that um leveraging generative AI for such things as better social engineering you know better fishing emails that are harder for human beings to detect you know based upon the you know ability to uh to develop um content that is is is more human like you know I think makes it so important in uh in today's environments to to really think um hard and fast about how you're going to um secure your data as you say which remains to be the life blood of most organizations right so you're really saying that it's that generative Ai and AI is sort of a mixed bag and that the the good folks have it and the bad folks have it and is there is is there someone who's winning that battle currently or is it hard to say you know I think I would say that uh the defense uh of the environment uh is is is probably ahead of the game you know certainly uh we continue to see um cyber attacks on the increase you know certainly you know I think that no matter how good your defenses are the challenge is is that bad actors will find a way um to often get through them and I think that's why uh we often talk about the the balance between preventative measures and measures to um allow you to recover when bad things happen you know so this concept of cyber security versus cyber resilience I think becomes even more important nowadays because the ability for Bad actors to to do things more quickly using generative AI uh techniques that maybe were much more manual in the past um now increases the speed that they can uh probe your defenses they can look across your uh your your it infrastructure for um places where you know maybe you haven't had the best of hygiene and you've you know left the back door open you know and the ability for actors to find those vulnerabilities you know is definitely an area of of opportunity uh for bad AES when they leverage generative AI so you know certainly we believe that you you really need to think about cyber resiliency you know as the the the posture that you really need to to put more effort into to make sure that the data that you rely upon is able to be recovered in a good known State whenever you may need to you know it's a really interesting idea cyber security is cyber resilience so in other words you a malicious actor has probed your your firewall and and maybe maybe entered um so You' you haven't necessarily blocked them but now you need to respond to them really well what does that look like in in real life what's an example of of how would you use AI to to repel that individual well it's not so much to repel I think one of the things that uh that the industry leverages AI today uh is for anomally detection you know and and you know quite often um you know many parts of the uh of the the the the the market from a vendor landscape perspective you know there's a lot of work that goes into identifying you know malware um identifying ransomware um and um really the position that we take is that rather than looking for specific you know bad code uh ransomware signatures we look at the uh the data that uh you have under management and looking at the backup of that data we look for anomalies you know we do um uh things like full content indexing and rather than looking for signatures of uh of of ransomware attacks or malware we're really looking for anomalies within the data itself anything that changes from a day-to-day basis one of the nice things about uh the backup and recovery process is that it happens on a regular basis so the ability to um uh compare good backups against Bad backups um identify differences in what you would assume to be similar data copies day after day or whatever period you're using from a backup cycle perspective allows uh solutions that leverage Ai and also machine learning to actually build a model of what good looks like and I think that becomes an important element about how Defenders and how vendors uh within the data protection space can actually identify anomalies to a very high degree of accuracy all right that fascinating point I think that really brings me to the next question you talked about this a bit what are the actionable steps you recommend to modernize data protection and recovery and it sounds like comparing backups using AI or ml but if you're going to give some real world advice to to modernize data protection and Recovery what are a few steps you'd say well that's certainly one uh you know I do think um when you think about modernizing data protection um when you're um investing in Enterprise AI infrastructure you know one of the things I think you need to start thinking about is treating your AI workloads as another critical workload within the environment whether or not you're taking advantage of AI infrastructure within the public cloud or what uh many large Enterprises are investigating is deploying AI infrastructure within their own Enterprise data centers and certainly um AI infrastructure has a number of elements you know you have the the training data the training um you know sources uh that you then uh leverage and feed into the AI models um and then you have the the prompts and responses that you um uh create and receive you know based upon the AI model that you've created all three of those levels of the AI infrastructure you need to think about how you're going to protect certainly a lot of the AI infrastructure uh that you tend to see in the market is very uh very much based upon a kubernetes container based infrastructure so certainly you know picking a data protection um offering that has a strong um capability to protect kubernetes you know is certainly uh a good um uh choice to make uh but then also U what you want is the ability to tailor your data protection policies and procedures to the AI infrastructure that you're using and one way to do that is to um pick a data protection solution that has a very strong um set of apis you know uh and give you the ability to uh to uh Drive the policies that you need for the specific AI infrastructure that you deploy by using a very much an API First Data protection solution in order to uh to build what you need the the API you know focusing on the API first approach does that also mean you can switch out uh Solutions easier because it's simply a question of plugging a new API or not necessarily you're not thinking that way uh I think I think often uh you know apis tend to be um specific to specific Solutions um so you know I think the order of the day is to do a good evaluation of um what you may be using and certainly you can imagine that most Enterprises um they're already doing backup and Recovery using you know a particular vendor that they may have been working with for a number of years the question is and one of the things that we see is that as the Enterprise infrastructure environment and the workloads that exist within that environment change then quite often that's when you you tend to get uh organizations uh invest time in looking at is the data protection solution that they're using today going to be um able to continue to protect all that they need to do as they go forward and especially as they start looking at um uh Enterprise AI infrastructure so certainly you know we've seen opportunities where we're having new conversations uh with uh clients as they start talking to us about we're interested in deploying Enterprise AI infrastructure um and you know one of the things that that we talk a lot about and it's a it's a phrase that um you know as emerging in the market we talk about the concept of an AI Factory which is really a um an umbrella term for the infrastructure that you need to deploy um uh AI infrastructure whether it be compute whether it be networking um whether it be storage um client technology but also data protection and certainly you know we see data protection as an integral part of an Enterprise AI deployment so you know that's where you're starting to see uh opportunities to have new conversations with clients about how to protect these new workloads you know that that I think is maybe the most central question and you you may have answered it and that that is I think so many companies are wondering I want to buy a solution but I'm not sure it's going to work now and 3 to 5 years from now how do I shop for that and figure out when I'm shopping how this is going to work you know two years from now and four years from now and it sound it sounds like your answer to the question is does it have that AI Factory approach it can do all those various things is is that your answer to the question like how how do I shop for something and know it's still going to be viable two to four years or now what what are The Telltale Clues I should look for yeah I mean I think uh you know some of it is is is looking at what data protection vendors have to say uh about uh their use of of AI um and uh their strategies for protecting AI you know I think you look at the data protection uh landscape and uh there's the ability to um bring generative AI into Data protection Solutions you know that's certainly you know one element of uh of then venders taking advantage themselves of generative AI but one of the things that we haven't seen a lot of discussions about by many vendors is when you deploy generative AI solutions for specific business outcomes how do you actually protect that environment the data that that environment is relying upon in order to to drive those business outcomes so you know I think you really need both sides of the equation you need vendors that are very active in uh bringing Ai and generative AI into their portfolio but you also need vendors that have a good handle on how to protect those new Ai workloads and I think that's something which um you know we've certainly been you know investing ourselves um you know with uh with with our new data protection offers you know I think the reality James is that um things are moving fast um you know AI infrastructure um is um more than likely going to create a uh large amount of new data um that all needs to be protected you know I think that uh that because of that your data protection solution needs to be able to handle both the the speed of of ch change uh and the scale uh that will come you know with these um AI uh workload and the AI infrastructure that that goes along with them makes perfect sense well I think that that all brings us to the question of the future of data protection AI you know what do you see in the next one to three years I'm wondering will we not get to a place where the infrastructure is basically safe because of the power of AI and generative AI I know that bad actors will always have it and good actors will always have it but I'm thinking back to for example your idea of uh examining the backups using Ai and ML and is is this backup you know still still in you know good and whole and whole I mean it seems like there would be so many ideas that that we could Implement with AI that eventually the infrastructure would actually really be safe and the idea of me malware at attacks would begin to fade away is that is that world ever going to come to pass or you don't you don't see it necessarily yeah unfortunately you know you know that I think is is um lines of you know we you know we see a future where you know we can always be 100% secure right you know I think we often talk about there's no such thing as absolute security you know I think that um you know one one area we haven't really touched upon is is um you know strategies around things like zero trust architectures right you know I think that that you're you're certainly seeing uh the industry and and uh organizations uh within uh the market um really look more closely at zero trust architectures to uh to make their whole environment including any of the things they do from an AI perspective uh more inherently secure but you know back to the point that um even with those strategies uh the question is can you ever be 100% sure that you're 100% secure and I think that that really goes back to what we started a discussion about is that you have to plan you have to plan for failure um and I think that um that's kind of like one of the the organizational models uh that I think is important is that you have to accept the fact that uh yes you want to do everything you can to keep the environment secure but you also have to plan for the um for the uh the time when um something happens that you need to recover and certainly part of the challenge you know we talk about things like social engineering we talk about things like malware there's so many different attack vectors that you really need to plan for uh the ability to always be able to recover your business whenever you need to you know and certainly plan is a a very key word there um you don't want to be thinking about recovery um uh in the moment you need to plan uh for the ability to recover you need to um rehearse the ability to recover uh and make it almost second nature so that when you need to use it you're not building a set of processes to get your business back back up and running and I think that's you know something which um is so important um with you know the the changing landscape that that companies are having to deal with planning and rehearsal there's often too little of that for sure uh Rob I think you said it a fascinating material you learned a lot it's going to be a really interesting sector to watch for sure uh I hope you come back and talk with us again sometime thanks a lot James

This transcript was generated automatically from the video's captions and may contain errors.