Latest / Now Innovating / Now Innovating: Cybersecurity in the Age of AI and Quantum Applications
Transcript
- 0:14Welcome to Now Innovating,
- 0:15a Research-to-Impact series at the University of Calgary.
- 0:18I'm your host, Julia MacGregor.
- 0:21This episode I speak with Dr. Hadis Karimipour,
- 0:23a Canada Research Chair in secure and resilient cyber-physical systems
- 0:27and an associate professor in the Schulich School of Engineering.
- 0:30We discuss cyber threats to critical infrastructure
- 0:33like utility centers, transportation systems
- 0:36and hospital networks,
- 0:37and the novel approaches
- 0:38her research is exploring to detect
- 0:39and prevent these cyber attacks.
- 0:43Hi Hadis, welcome to Now Innovating.
- 0:45Hi. Thanks for the invitation.
- 0:46I'm so happy to have you here.
- 0:48I'm really excited to like,
- 0:49learn more about your area of expertise,
- 0:50which is like cybersecurity
- 0:51and resilient systems.
- 0:52My reference... like cyber threats related to digital systems
- 0:55I think of like phishing attempts,
- 0:56like trying to get my bank information.
- 0:58And steal my personal information to.
- 0:59But could you share some examples for our audience
- 1:01really of like cyber-physical systems
- 1:03and, you know, the possible cyber threats
- 1:05that are facing them.
- 1:05So let me begin with some definition
- 1:07of the cyber-physical system.
- 1:09What we call a cyber-physical system is the integration
- 1:13of the digital technology and physical world.
- 1:16So these are vital systems that we rely on
- 1:19for our daily life,
- 1:20like energy centers, utility centers
- 1:23that deliver electricity to our home
- 1:25transportation system
- 1:27that moves peoples and goods
- 1:29or hospital networks that’s obviously saving lots of lives.
- 1:33So these are vital and very important systems.
- 1:35But at the same time,
- 1:36they are very vulnerable
- 1:38and sensitive to cyber attack.
- 1:40What is the reason?
- 1:41The reason is that most of these
- 1:43critical infrastructure
- 1:44they were designed like 100 years ago,
- 1:47with the assumption in mind
- 1:49that they will work
- 1:50in an isolated environment.
- 1:52They're not going to be connected
- 1:53to... in different systems.
- 1:54They're not going to be connected to internet.
- 1:56So with these assumptions,
- 1:58these systems were built,
- 1:59but right now we're jumping into digitalization.
- 2:02We are connecting all of them together to the internet
- 2:05and Internet of Things devices.
- 2:07And that makes them like a playground
- 2:10with lots of vulnerability
- 2:11for the cyber criminal
- 2:13to take advantage of it and launch
- 2:15different types of cyber attack.
- 2:17And we see lots of example, actually, in the real world.
- 2:20For example, for energy centres
- 2:23we can see examples of false data injection attack
- 2:26when cybercriminals
- 2:27change their readings of the sensors
- 2:29and provide misleading information
- 2:31for the operator,
- 2:32and that will result in mismanaging
- 2:34the electricity flow, for example,
- 2:36and in, in severe cases like Ukraine cyber attack.
- 2:41You can see widespread
- 2:42blackouts and power outage.
- 2:45And the same with
- 2:45for transportation system, for example,
- 2:47we have lots of smart traffic lights.
- 2:50We have autonomous vehicles, we have,
- 2:53automated railway control
- 2:55and all of these,
- 2:56if they are readings of the sensor
- 2:58and measurement are attacked,
- 3:00by integrity attacks,
- 3:01denial of service attack,
- 3:03any other type of attack
- 3:04that can result in falsifying information
- 3:07that may cause accidents,
- 3:08block emergency services
- 3:10and lots of other,
- 3:11basically, damages to the system.
- 3:13Oh, wow. So as we're making ourselves more efficient,
- 3:14we're really just opening ourselves
- 3:16up to more vulnerable vulnerabilities.
- 3:17I think really.
- 3:18We are getting lots of advantages,
- 3:21for moving toward
- 3:22this cyber physical system
- 3:23and in this digital era.
- 3:25But at the same time,
- 3:26it will bring lots of new
- 3:28challenges and risks for us.
- 3:30So we have to make sure
- 3:31that we other skills challenge us to be able
- 3:33to take advantage of the technology.
- 3:35Thank you for that context and background.
- 3:36I really appreciate that for my own awareness to.
- 3:38Where do these cyber criminals, these cyber threats,
- 3:41where are they mainly coming from?
- 3:42Like, and how are they detected?
- 3:44They are coming from a variety of sources actually,
- 3:47if I wanted to categorize them
- 3:49to three main categories,
- 3:50I would say
- 3:51the first one would be a nation state
- 3:54or a sponsor type of attacker
- 3:56who are usually targeting
- 3:58critical infrastructure for a strategic gain.
- 4:01These are groups that are backed
- 4:02by the government support.
- 4:04So they are highly sophisticated
- 4:06and very well equipped.
- 4:07And that makes them very difficult
- 4:09basically to defend against them
- 4:12and protect ourselves against them actually.
- 4:15The other group
- 4:16would be organized cyber criminals
- 4:19who are mainly motivated by financial gain.
- 4:22And they are working like, professional enterprises
- 4:26with huge expertise and skills,
- 4:29and they usually rely
- 4:30on the black market tools.
- 4:32And the last what I would say are insider threat,
- 4:35they pose as unique risks
- 4:38because of their legitimate access to the system
- 4:40and the level of sophistication can vary.
- 4:43It could be a like just a negligent employee
- 4:46who mistakenly cause a vulnerability,
- 4:48and a cyber attack opened a door
- 4:51for a cyber attack in the system,
- 4:52or could be an angry, employee,
- 4:55an insider threat that leverages
- 4:58all those unique knowledges
- 5:00to run a high profile
- 5:01and basically an attack
- 5:03with a huge impact on the system.
- 5:05So we have different,
- 5:07groups of the attacker
- 5:08with different level of sophistication.
- 5:10Traditionally,
- 5:12how we defend them
- 5:13was mainly relying on IT technology,
- 5:17relying on network traffic, analyzing IT
- 5:20login information like this.
- 5:22But nowadays modern cyber security
- 5:25mainly relies in a layered approach.
- 5:28And integrated threat intelligence,
- 5:31behavioral analytics,
- 5:32dynamic monitoring and things like that.
- 5:36For instance, with AI,
- 5:38we are able to see lots of complicated
- 5:41and complex patterns in the system.
- 5:43You can analyze them in detail,
- 5:45and that will help us to know
- 5:46about, possibility of the cyber attack
- 5:49a lot faster than traditional tools.
- 5:51Or when we see a problem
- 5:53from different angle in a multi-layer
- 5:55basically approach,
- 5:56we can see changes in the digital
- 5:59basically level of the system,
- 6:00and at the same time changes in the physical,
- 6:03the fluctuation of the voltage.
- 6:05So that will give you excellent
- 6:06insight about possibility of the attack,
- 6:09where it happened,
- 6:09how it happened,
- 6:10from which group it's coming.
- 6:12And when we use threat intelligence,
- 6:14which means different organizations
- 6:16and governments, industry, collaborate to share
- 6:20all the available data about the possible threat
- 6:23group of attackers and things like that.
- 6:26That, again, gives us extra source of information
- 6:29to be prepared to defend against the attacker.
- 6:32So combination of all of these,
- 6:34is the approach that most of the organizations
- 6:37take these days.
- 6:39But how effective are these?
- 6:40It totally depends on the level
- 6:42of the sophistication of the attacker
- 6:44and the defense mechanism.
- 6:46Oh my goodness. There's something I didn't realize.
- 6:48There's like three categories of cyber criminals.
- 6:49Mainly you can categorize them.
- 6:51But if you want to break it down
- 6:52you can go to lots of other category.
- 6:54But broadly it can be categorized.
- 6:57Where does the dark web fit into this.
- 6:59Yeah. The dark web is the basically the platform,
- 7:02let's say, that provides lots of tools and,
- 7:07algorithms that cybercriminals
- 7:09maybe buy or sell, for example, when, most of the time
- 7:13when there are a group of cybercriminals
- 7:15that launch a ransomware attack,
- 7:17they get the data and they threaten the, basically
- 7:21the people that we will, sell these data in the dark there,
- 7:25and there are lots of other criminal groups
- 7:27that might have better resources
- 7:28or better motivation to get access to those data
- 7:32and misuse those data.
- 7:33So it's a platform for people
- 7:34that illegally share information.
- 7:37It could be software, could be data,
- 7:38could be different tools.
- 7:40And any things that, you cannot buy
- 7:42legally in, other websites or platforms.
- 7:45So I was really excited that we did
- 7:46have the opportunity to visit your lab, the,
- 7:48smart cyber physical lab.
- 7:49And we heard from your graduate students,
- 7:51your postdocs, about the research
- 7:52that they're doing under your direction.
- 7:54I would love now if you could tell us
- 7:55more about the novel approaches
- 7:56that you're exploring,
- 7:57to, you know, prevent cyber attacks.
- 7:58In my team, we are, taking several different,
- 8:01novel approaches. And I will, explain,
- 8:05a couple of them here, just generally not too technical.
- 8:08One of them would be what I call a,
- 8:10AI driven, defense mechanism.
- 8:14Predictive defense, actually.
- 8:15And, how it works traditionally,
- 8:18cyber
- 8:19security tools raise an alarm
- 8:21after an attack happened. The system.
- 8:23But what we will do in our approach
- 8:26is that we are flipping that basically, technique.
- 8:29And we wanted to predict the attack
- 8:31before it happens in the system.
- 8:33It's kind of like the same things
- 8:35that we do with the weather forecast.
- 8:37We use wind data, temperature, humidity
- 8:40and all this information
- 8:41to predict possibility of a storm.
- 8:43So the same way we employ different AI
- 8:46and we develop different AI algorithm
- 8:48to analyze patterns
- 8:50of any suspicious activity
- 8:52activity in the system dynamically.
- 8:54And that will give us the information
- 8:56and insight to predict a possibility
- 8:59of a cyber storm in the system
- 9:01so that way we can save time to, basically
- 9:05take care of any vulnerability
- 9:06in the system or be prepared
- 9:08if the attack happens
- 9:09so we can isolate it faster and
- 9:11prevent propagation in the system.
- 9:13That's really cool.
- 9:13That makes me think of the movie Minority Report...
- 9:15seeing the future of the attack!
- 9:16Yeah, exactly.
- 9:17So the other approach is,
- 9:19quantum resilient cyber security.
- 9:22We know that in the quantum era,
- 9:24with the power of the quantum computer,
- 9:26all the a strong encryption key can be,
- 9:29basically can break in minutes.
- 9:31And that means that
- 9:32it will pose
- 9:33huge risks to financial institution,
- 9:36critical infrastructure,
- 9:37energy grids and many more systems.
- 9:41It's like that right now.
- 9:42Somehow, with all the security tools
- 9:44that we have, we have a wall with,
- 9:46with strong encryption tools that we have,
- 9:48we have a wall in front of cyber criminals,
- 9:50so not everyone can pass through it
- 9:53and come in, attack the systems,
- 9:55but with it, when encryption is gone,
- 9:57we don't have that wall,
- 9:58so we will have huge number of the attack
- 10:00that makes it more difficult to deal with,
- 10:02even with strong AI tools.
- 10:05So what we are doing in my research is that we're
- 10:08also trying to ensure that our cyber security tools
- 10:11are resilient, in the quantum era.
- 10:15The other approach that we take
- 10:16is that we study to see how,
- 10:19quantum items like entanglement, superposition
- 10:23and things like that can result
- 10:25in subtle changes in the data
- 10:28that can trick an AI model,
- 10:30because we rely on AI for cybersecurity.
- 10:32So if tricking AI model is easy in the quantum era,
- 10:35that means we cannot trust the AI
- 10:37and that that creates a world of chaos.
- 10:39So we use those insights to ensure
- 10:41about the security of the AI model as well.
- 10:44And with these combinations
- 10:47we are trying to ensure
- 10:48that we are safe and secure.
- 10:50Also in the quantum era.
- 10:51And finally, the other thing that we consider,
- 10:54is that we will consider the human dimension
- 10:58of the cybersecurity.
- 11:00Most of the time, people focus in the technology.
- 11:03It's like we just have to develop a strong technology.
- 11:06But humans are always an important element
- 11:08and aspect in cybersecurity.
- 11:10They can make mistakes under stress, fatigue
- 11:14or when they are misled by misinformation.
- 11:16And cyber criminals know that
- 11:18and they exploit it for, phishing campaigns - developing
- 11:23different phishing campaign and email,
- 11:25through deep fake profile, social engineering
- 11:28and different things like that.
- 11:30So what we do is that we take advantage
- 11:32of cognitive AI tools that analyze human behavior,
- 11:36and we are trying to see how we can helps
- 11:39people under stress and under pressure
- 11:42to make sure that our tools can, detect
- 11:45signs of, stress, fatigue and things like that,
- 11:48or can adaptively support people under
- 11:51stress and pressure to avoid mistakes
- 11:53during the security call scenario,
- 11:55like under cyber attack.
- 11:56So by considering this human dimension,
- 11:58we ensure that our solution
- 12:00is more complete because,
- 12:02we believe that cyber security
- 12:03is not just about having a strong technology, it,
- 12:06but it's about protecting
- 12:09and safeguarding the people's
- 12:11and gaining the trust of the people
- 12:12who rely on those technology.
- 12:14So these are
- 12:15a few examples
- 12:16of the approach that we take.
- 12:17So generally we believe
- 12:18that we should not only,
- 12:21detect and attack, we have to expect it.
- 12:24We have to predict it.
- 12:25We have to protect our systems
- 12:27against technology of tomorrow,
- 12:29and we have to ensure that we gain people
- 12:31trust and support, people who are basically reliant
- 12:35on our technology.
- 12:37Like, those approaches are fantastic.
- 12:38That's so cool.
- 12:39Like you're doing all that work...
- 12:41You're predicting, using quantum
- 12:43I didn’t realize you’re using quantum in cybersecurity,
- 12:45I guess, like with those approaches you're doing.
- 12:47Can you tell us about impact that your work has had on
- 12:49maybe cybersecurity policy in Canada or beyond?
- 12:52Honestly, this is one of the important thing.
- 12:54And something that I value a lot for my work
- 12:56to make sure that we make a bridge
- 12:58between research in the lab
- 13:00and real world impact.
- 13:01And we try lots of different ways
- 13:04to ensure that by participating in different events,
- 13:07showcasing our, basically, achievements
- 13:10and tools in different environment
- 13:13to ensure that it reaches a broad audience.
- 13:15And, we see different basically signs that shows
- 13:18we were successful so far.
- 13:20For example,
- 13:20several of our publications have been cited
- 13:24by the European Commission and votebank,
- 13:27which shows the impact of the work
- 13:28in national and global policymaking.
- 13:31And I also personally always participate
- 13:33in different, committees like the standard Council of Canada
- 13:37and other policymaking committees.
- 13:40And I try to ensure that,
- 13:42what we are developing and deploying is
- 13:45used in a responsible and, ethical
- 13:49and internationally interoperable way.
- 13:52And on top of it,
- 13:55we also value a lot commercialization.
- 13:58And, I always recommend to my students
- 14:00when they wanted to pick a topic
- 14:02for their research, I always recommend them
- 14:04to go through different,
- 14:05maybe industry, and search to see
- 14:08what are the challenges that they are facing,
- 14:10maybe interview with them.
- 14:11Just look at the website
- 14:13and any data that's available
- 14:14so you can pick something that, will help
- 14:17the actual, basically, problem in the real world.
- 14:21And different people in the team are working in
- 14:24commercializing different tools
- 14:25and, platforms that we are developing different
- 14:29AI algorithms, AI models.
- 14:31If I wanted to mention examples,
- 14:33it would be a quantum AI cybersecurity
- 14:36platform that we are developing
- 14:38for critical infrastructure.
- 14:40and ARTIsync which is a platform
- 14:42we are working on for the intervention,
- 14:45AI based intervention planning.
- 14:47And, on top of this effort
- 14:49that we take toward commercialization,
- 14:51we are also working
- 14:52with different industry
- 14:53partners, from utility sector to insurance firms.
- 14:57And we, try to make sure that what we are developing
- 15:01as AI based cybersecurity tools, will be implemented
- 15:04in the real operational environment.
- 15:06So with this dual, kind of commercialization
- 15:10and industry collaboration,
- 15:12we are trying to ensure that all work
- 15:14has both societal impact
- 15:16through increasing resilience of the cyber ecosystem
- 15:19and economic impact
- 15:21by bringing innovation to the markets.
- 15:23I love when you said that you're
- 15:24focusing your research-to-impact
- 15:25like that's what this show is all about.
- 15:27So I'm so happy that you're a guest on here and sharing all this fantastic
- 15:29that work you're doing. And that's really cool
- 15:31with the commercialization piece.
- 15:32I didn't know you're working on that.
- 15:33Yeah, it's really difficult actually.
- 15:34That's one of the challenge
- 15:35when it comes to research,
- 15:36because we don't have
- 15:37that much of funding to focus
- 15:38on the commercialization.
- 15:40And, we always rely on the funding.
- 15:43But, recently we were, successful
- 15:45to get, several sources that support us,
- 15:48to commercialize what we do in the lab.
- 15:50So we're using those resources to,
- 15:52take our product to the market.
- 15:54As you said too, like, with your research being used
- 15:57on European policy inside of that
- 15:58or on your own,
- 15:59like work on policy and committees
- 16:01like you've been
- 16:01recognized internationally, really,
- 16:02for your expertise in cybersecurity
- 16:04and you’re a Canada Research Chair in Secure and Resilient Cyber Physical Systems.
- 16:08So how do you foresee the global cyber security
- 16:12landscape kind of, you know, changing as you're talking about,
- 16:15you know, advancements in like, AI is being hacked and tricked
- 16:18and like with quantum applications and machine learning, like,
- 16:20how do you see this evolving as it goes forward?
- 16:23I believe that we cannot ignore
- 16:24the fact that technology like AI
- 16:27are changing the cybersecurity landscape.
- 16:30But I believe that I specifically
- 16:33is the double edged sword
- 16:34on the defensive side, it really gives us huge powers
- 16:38to, react against the attack, to protect ourselves,
- 16:43to neutralize attack, to develop systems
- 16:46that are, automated, self-healing,
- 16:50to have infrastructure that can detect the attacks
- 16:53and see it's predicted, and then,
- 16:55regain their normal operation.
- 16:57But on the offensive side also,
- 16:59the attacker won't start to steal.
- 17:01They take advantage of the AI to develop,
- 17:04highly, personalized, phishing campaign, deepfake profile,
- 17:10malware that can, basically dynamically change
- 17:15and change their features.
- 17:16So they can easily evade traditional systems.
- 17:19So what we see would be a battle of AI against a lot.
- 17:23But again, I believe that future of the cyber
- 17:26security won't, only be owned by AI.
- 17:30If we only focus in AI, we will basically suffer a lot.
- 17:35What I think is that we have to.
- 17:38We change the way that we design
- 17:40and manage the system so it can start with, having architecture
- 17:45like zero trust architecture
- 17:46when we don't trust anything in the system,
- 17:48any users, any systems, any component in the system,
- 17:52you want to to authenticate everything.
- 17:54So what we have to do is that we have to ensure that cyber security
- 17:58is considered in the design step.
- 18:01It should not be an afterthought.
- 18:02What's happening right now
- 18:03is that we design something.
- 18:05And then we wanted to add cyber
- 18:06security component or feature to it,
- 18:08but it's not going to work. Yeah.
- 18:10We're going backwards. - Yeah. Exactly.
- 18:12And at the same time
- 18:12we have to ensure that the system
- 18:14will be designed resilient.
- 18:17It would they would be able to recover
- 18:20from the attack and gain
- 18:21back to their normal operation.
- 18:23That that's also important.
- 18:25And ultimately I think the global winner would be,
- 18:28those, people that combine innovation with policy,
- 18:34governance, ethics and everything,
- 18:36that that's the only way
- 18:37that we can make sure
- 18:38that we have, a digital future
- 18:41that is safe and secure. So.
- 18:43And if I wanted to phrase it differently,
- 18:45I would say cyber security is not
- 18:48only protecting the machine,
- 18:50it's protecting people's
- 18:52trust and safeguarding people's
- 18:54trust in the system
- 18:55that they rely on in their daily life.
- 18:58If we ensure that we design
- 19:00with innovation, responsibility,
- 19:02collaboration and, basically ethics
- 19:05in our mind and resilience,
- 19:08that's where you can have a safe
- 19:09and secure digital future.
- 19:10I love that. That's a great note to end on.
- 19:12Thank you Hadis so much for joining me today.
- 19:14It's been an absolute pleasure
- 19:15learning more about your work.
- 19:16Thank you very much for the invitation.
- 19:18I really enjoyed being here.
- 19:20It's astounding the importance of cybersecurity
- 19:22and the efforts to protect critical infrastructure
- 19:24and advanced policy in this space.
- 19:27While we still need to be cognizant of our ever
- 19:28growing digital worlds
- 19:30and efforts to protect ourselves
- 19:31at home from cyber criminals.
- 19:33You can find links in the episode description to learn
- 19:35more about Hadis’s research,
- 19:37and to learn how to move your research-to-impact
- 19:39in the UCalgary Innovation Ecosystem,
- 19:41visit ucalgary.ca/innovation