Balancing AI Innovation with Data Protection: Risks and Controls: Pooja Shah and Atilade Ayanbadejo

uh hello everybody my name is a um I'm a second year Master student at the Concordia University of Edmonton and um yeah I have specialized experience in Lino system administration um experience in Cloud security and information systems um management so our um this is Puja can you introduce yourself too hello everyone my name is Puja and I've been into information security for about 8 years now but I feel like you know there's still a lot to learn I think everybody feels the same way right okay so first of all we are very honored to be here presenting in front of all of you we have you know gathered some amazing support and guidance from the

Edmonton cyber security community so we are really grateful and without further Ado let's just dive into it okay so um our top topic is um balancing AI Innovation with data security and risks and controls so I'm going to run through a brief outline um I'm going to give a brief introduction okay not so brief but then introduction and we're going to discuss AI tools across different Industries making impact across different Industries and we're also going to discuss some of the statistics to prove some of our claims and discuss some risks some security challenges and also some controls so yeah in the so like in this age of artificial intelligence our data has become more valuable than ever you know

like every click every purchase and you know like us from our browsing history everything is like a digital footprint of ourselves yeah so like while these tools have the potential to revolutionize how we work like how we do things like it's also a double-edged sword so a prime example is the rise of deep fakes so like you ask yourself what is a deep fake so a deep fake is a synthesized video or manipulated video that that convincingly project somebody doing something they actually didn't do and you know it blurs the line between fact and fiction so um so like today you know as individuals as organizations as we Embark as we embark on this journey to

navigate AI in this in this AI driven World we're going to look at the potential benefits and the perils we're also going to look at the we're going to weigh the good and the bad side of this technology so and don't worry we're not going to leave you in uncertainty our mission is clear to equip you with the knowledge and controls to empower that empowers you to take control of your digital identity so like why do we need to protect our data in the context of AI so as individuals as we immerse ourselves into this AI driven world right like our data has become very very invaluable like and this not just like before where

it was like Pi like um personal identifiable information where you have your location in the database your name your age but now we're having information about your behavioral analytics like how you click every like the people you call every day that specific time of the day where you text somebody like I don't know if you use your Apple phone and sometimes he says okay you can usually call this person at this time during this day so like those that information is just not okay I have this number I have this information about myself it's telling me how I behave over time and you know this all these advancements they drive the AI algorithm like they drive us and like it

also it also like it also brings about the topic like converging AI with data security it poses questions like questions it poses questions about like ethics privacy and security okay so like also The Duality of a AI you know like in this Dynamic landscape of AI tools you know while while there's also a potential for um there a potential for Ben that benefits us there's also a potential for data exposure so no why AI promises to revolutionize Our Lives you know through empowering us with um productivity just like you know like um Siri like on your phone like hey can hey Siri can you set a reminder for me by 7 p.m. like you know all these things that AI can do for

you ask TR oh can you make an itinary for me like I'm somebody that loves going to the gym a lot like can you make a meal plan for me and like it does all this like the same information that does all of that like the same technology that does all of that it also exposes our most sensitive information to unprecedented risks you know these risks are exposed to all the time because of okay like everybody just says like a is very very easy to use oh I just use it to create whatever but like all this information we given it like we also have to think about the risk and like the privacy issues that we face with it

so like I'm going to talk about some like I'm going to talk about like AI tools across Industries cuz I just want to give like since like AI has been there before like different companies Google Apple has been using it before but now it has been released to the consumers like tools like Char GPT me joing I'm just going to discuss some of them some Cutting Edge tools making transformative impacts in different organization and different sectors of organiz ation so yeah in different Industries we have like the first industry I want to talk about is data processing and Analysis like you know these tools can help businesses to like automate tax like increase productivity analyze data like a perfect example of

this tool is um examples is Tablo Microsoft power beer and altox so like Tableau what does it do it takes raw data and turns it into dashboards like and reports and all this information that is getting like it's raw data from different aspects of business organizations so like what is the impact on data security like potential for data breaches as usual like all this information like that is being released like if it falls into the wrong hands like they not just getting information just about oursel but then like processed information like things that have to do with our analysis also like another industry that is receiving like cottonage um taking advantage of this cuton Edge tool is software development

and testing you know like ai ai can be used to like automate code generation code testing code like different things can make it easier you can easily use charp to create a Lander code okay P can you help me create this specific script to do this specific thing so like some of the specific tools we use are Char GPT GitHub co-pilot and catalon and the impact of the um data security I would say is like like imagine if the code was insecure like there's a risk of insecure code another industry is the customer support and engagement industry like most of us right now when we go to sit going to are the days when we had to

like talk to somebody in the customer in the chat like the chat B nowadays are more easier to use like you just need to they are frequently asked questions that people always ask all the time so like why do you have to put a human being there when you can just easily put a bot that's going to ask you some of like it's going to answer some of those your frequently asked questions so there's an issue with also data privacy because the information you are giving it like where is it being stored and like that's the essence of this our presentation is it's not just giving you the risk it's not just about using this AI tools about

like what is our information being used for like everything about privacy is like okay can I use like why do we care about privacy so much like can I use this and not really be afraid that somebody's going to use this data against me or you know sometimes when you're like oh oh you're just talking to somebody I'm like oh I just I want to get new shoes today and like you now see on your Google and YouTube you start seeing ads about new shoes I'm like wait are these guys listening to me like what's happening here like who are are they using and nobody anytime I asked that question like nobody has ever given

me a straight answer like what are the laws what are the Privacy Law what what who's using our information so like we also have voice assistants like um that industry too like a lot of people I know they use Siri um Google Assistant I use Siri a lot like whenever I want to set reminders I just say okay I hope your like your phone is muted but then like hey Siri can you set a reminder for me to go to class or have I have a meeting on next week Tuesday 7:00 p.m. at this location so first I have my location I have the time I have where I'm going like I have different information in

that prompt so like that information where is it being stored like everybody trust Apple but like really if there's a breach this specific information can be used to do um personalize attacks to you so we also have okay we have cyber security and risk management and we also have marketing and sales you know there are spec and one of the popular tools out there is mid Journey like I'm sure some people have used it I've used it a lot like you know g at the days when you like when you in my previous workplace like if you had to create a cyber security awareness training like you have to create an infographic um PowerPoint but then

nowadays like M joury all you have to do is okay I want to create a cyber security awareness training can you create a specific picture maybe with like you know you want to create a fishing campaign or you want to create a fishing slide all you have to do is just tell it okay create a hook with maybe a mail on it and you get the exact picture so you don't have to search like before it creates what you want like you can create what you want from so it's a very powerful tool but also issues of like um copyright infringements and all those things like P is going to talk about that in the second part but then these

are some of the issues like this is like this is where a is being used in Industries and really making transformational differences but then yeah there like I also like to like talk about some statistics to prove my point to prove the adoption of a in different Industries and like um first statistics this from fobbs and it says the AI Market is projected to reach a staggering 47 billion by 2027 experiencing substantial growth from his estimated 86.9 billion look that's over 300% increase like it just shows that even we we thought we were ready for AI but it's just showing the adoption of the adoption is really really increasing in consumer space you can also see Char

GPT had 1 million users within the first 5 days of being available like you always wonder like what kind of app will have 1 million users like how useful is that app that it gets 1 million users in the first 5 days like productivity app that is being used by I know most people here in this room like you can't tell me you're not using Char GPT but it's all right and then it is expected that 10% of vehicles will be drivered by 2030 and you know like electric vehicles are the new thing now everybody wants to have an Evan like self-driving Vehicles who is going to be in charge artificial intelligence like when you turn left

turn right making those decisions okay maybe you have to choose between hitting a pedestrian or hitting a car like who's who makes those decisions those so AI is going to be part of it just shows like how how many companies are taking it Taking um and you see also voice search is on the r with 50% of users using your daily and like that's something very personal to me because I know that like while doing my research I saw some stories of voice clone scams where people use your voice and like one thing about having your voice this are like sometimes you you think you are unique because I only me has my voice but

imagine you recording my voice and putting my voice through software and then replicating that voice and then calling maybe my parents and telling them oh I'm in trouble I need you to send money to me like these are examples that can be used with um voice like this this information that we easily given oh to make our life so easy and can also be used to destroy us and like you know in also in countries and like um n nationalities like who is leading the adoption and he says China leads in adoption with 58% of companies deploying Ai and it just shows that yeah China is not slowing down like they know this is going to really improve productivity

going to improve impr profit it's going to improve like even like people that want to go into like military Warfare like this is this tool is going to help it and also like it shows the most common ways users plan to use artificial intelligence like you know I was talking to somebody and they say most people say they use it to craft emails like you know plan travel itinary somewhere like me us to plan my um gym plans like meal plans and you know like but all this information is personal information and like you know when you're putting it you might do like oh can you create a sample information but then when it's not

giving you answer you want you just say come on I put all my personal information just to give me what I really want and you know this this just shows that okay yes there's adoption but there's also a lot of people a lot of people don't know what the risk of what they are doing with this tool and what what are the laws and regulations so yeah those are some of the statistics but then I want to just give a brief overview of some of the risks that we may we face with these two and the first R is the existential risks and there's something called artificial in general intelligence I you know while we are not close to artificial general

intelligence like there's the possibility of AI like overtaking human intelligence and it also it it needs to be monitored like you know those days we watch all these Sci-Fi movies like I always see most Sci-Fi movies I see the reality right now like it just begs the question like oh my God like will will this will this be the end of the world and like um another risk is biased algorithms and like you know AI is always trained on data like its main catalyst is data you know there's a saying that like if a lie is told long enough like everybody is going to believe that light so like what's to say that this data is not

trained on wrong data like what's the bias that okay this thing is this specific way like I can just start putting so much like doing our research like there was this person the grandfather of AI I forgot he used to work in Google he says that like when you want when you wanted to train when they were first training AI maybe they wanted to train model to say okay this is a fish so in order for that model to understand it like they had to put so many pictures of fishes like so much data of fish so that by the time you see you just see even a tail or a f you already know that this is the fish so

like imagine putting so much wrong data out there into this AI algorithm like and people now start asking and sometimes I'm sure you had this issue when you use this tool and like when you get you ask it for it gives you a wrong answer so confidently right like you think it's like this is so wrong but it's so right at the same time because of the structure so like that's something about biased algorithms and we also have job displacing like everybody says yeah there will be new jobs but let's not forget about the people that don't know about AI because like just like how all of us say have specialized knowledge like there are still a lot of

people that don't know about Ai and there's going to be that Advantage you have over them so a lot of people a lot of jobs that require repetition will be lost and the new jobs that will be created are not just going to they're going to be people of people who who require who know how to use these tools and data privacy personal data exploitation like you know one thing I always talk about is like our data that's like I hate the fact that whenever I say something or I say I'm looking for something I start seeing I know sometimes I start seeing ads about those things but then what is my data being okay is it because like is my

microphone being used without me being like I don't mind people giving me an easier way to get a specific thing but I want to know that okay I gave out that data like it was my under my own permission so like there are data privacy issues and you know intellectual property as we spoke about like mid Journey like people you know there was a time when Del came out a specific Ai and everybody was just creating like different specific Arts like malalas of themselves but like the where where was this AI taking his data from where was his Source from so like this are there any laws protecting against artist against um people that create such such

um um such arts and PR will talk more on that and we also have misinformation influence campaigns where people can just use AI because of his ability to scale to create have on social media like creating because of the amount of data it has it can just create so much misinformation out there and we also have ai scans like I spoke about like voice cloning like you know deep fakes identity fraud like just I still feel like this is still the one of the biggest scams out there because like there's a thin line between fact and fiction like you can't really know if you see a video of me out there saying der derogatory wordss or bad things like

first thing you're going to do is believe before like you're going to nobody's going to verify if it's me because it looks exactly like me so that's a very big big risk and concern and you know we have potential to potential to impact at a personal level like you know we don't know the future but I feel like they're going to be like you can call me crazy but I think they're going to be robots in the future and like this robs are going to be based on AI and like a lot of people are going to have relationships with this like they can have relationship that's like maybe like a like you know maybe like

Jarvis in Iron Man like and over time like when you have ai telling you like listening to you every time and like telling you the right things all the time and like you're always having conversations going to affect our conversations with other people because like we are always used to getting yes hearing everything right all the time but then when we now meet other human beings who are Dynamic who can say no too like we are just like it's going to affect our personal relationships that's what I feel and then like like we have offensive cyber attacks like more targeted cyber attacks there's a social economic Gap like people that don't know about AI like and people that know about

AI there's a big gap and you see that big gap like it's like yes it's unfair but it's true like it's just like like you having a special skill specialized skill gives you that Gap against somebody that doesn't have it that's just used to it and it's going to the Gap is already there but it's going to increase and we also have National Security like you know for countries countries like nation states planning Specialized or targeted cyber attacks to PE to countries like countries need to protect themselves they need to create laws like this is not just a company thing it's not just an organization thing like it's a world thing and that's where Puja is going to explain more from

our own point of view based on the laws the ethics and how government and organizations can align themselves to create a peaceful world so yeah I'm going to give it to Puja now thanks yeah he has already set the stage for me so I think many here would agree that artificial intelligence is a revolutionary technology so with a quick show of hands can you tell me who of you are excited about the techn about our future with eii and who's concerned about it very understandable we have an interesting mix and and it's okay to have any kind of approach that you choose because right now we don't know where we are heading right so if you ask me well looking at all the

different possibilities and potentials of AI I am really excited to see the future but let's be honest it's not just AI this Quantum Computing cryptocurrencies metaverse machine learning and so many more Technologies which are advancing quickly too all of them together can do wonders or can shatter us completely and it's very difficult for anyone to predict at this point about the outcomes positive or negative and also how far are we from this complete transformation sorry about it thank you so anyways we will just start with the first point now so uh um but one thing is clear we need to manage the risks so that we can create a safe future where we can trust

and maximize the benefits of this technology Ellen musk has repeatedly stated that AI has the potential to destroy entire civilization it is a pretty huge claim and when the risk is too high it needs to be managed at all possible levels it is crucial that we all pull together to manage the risk and this V are these four broad categories first is government it is the first time in history where Tech leaders are actively approaching the government calling for an action asking to regulate them this is a unique situation and effective regulations can surely curve misuses of AI especially by the malicious actors next is a AI companies who better understand the nuances and and intricacies of AI than

those who built it so they have a significant responsibility in driving and guiding the controls of AI and last it's us so we are at the uh receiving and of the impact but at the same time we also have the power to to influence its trajectory especially us working in cyber security and related tax tax so these are the next two um stakeholders businesses and the individuals businesses they have more responsibility many of you might be using AI for your work so you have an added responsibility when you are using AI so in the next few slides we we will explore more specific about each of the player more specific controls about each of the player and these controls are

drawn from expert insights research and recommendations from leaders so let's dive into the first one yes so government government must deeply understand Ai and its implications to design the regulations they can accelerate this process by collaborating with different Tech leaders scientists researchers and maybe people from some other sectors like ethics they can give their insights and they can make this process faster some of the com uh countries including us have already embarked on this journey as we all know most governments are still struggling with postco recovery economic challenges so keeping us steady Focus can be challenging which brings us to our first control strategic planning government should at least uh plan at the earliest about their road

map the budget and if necessary may be a whole separate agency to exclusively focus on AI government should also consider having visibility over ai's Progressive cap capabilities and should impose Advanced controls over the AI models that possesses high risk or maybe the ones that c to a vast user base that's there's another domain that we want to discuss is legal framework and accountability so first we have data privacy that's a burning topic so we took years to understand the complexities of data privacy in the in the era of social media and it is high time now that governments step in and they create some robust framework where uh data security is part of an ecosystem

and not an afterthought recently many artists and creatives have expressed their concerns about ai's involvement in copyright issues including the creators for the famous show Game of Thrones in cases like this the creators must have an autonomy to decide if their work is used to train the AI systems now if anyone is thinking how practical this is there's an entire debate on that what about transparency and public trust so just think about it like a nutritional label but for AI imagine a scenario where AI tools come with a scorecard where they show their data diversity transparency user engagement it can give you some some insights about the platform and then you can take an informed

decision last is user education and risk management so we will talk more about it later but right now just make a note that some of these controls might overlap with multiple players now uh I think we will not have enough time to go through the regulations in detail but but the gist of it the gist of it is we need kind of a global collaboration there are so many initiatives and discussions happening all over the world but to tackle this techn huge technological challenge we need effective Global collaboration and this will prevent redundant efforts it will it will reduce time for us to get to the results and it will promote better understanding to tackle the cross

border threats okay so now let's talk about the AI companies they are the two true drivers of this AI vehicle they can direct us in the right or wrong direction for AI companies ethics play a key role especially where the boundaries are still being drawn it's all about the proactive steps that the AI companies can take especially towards security and ethics so this brings us to our first point ethics and inclusion AI companies have to maintain a very high ethical standards and promote a broad vision of inclusion it is important to ensure the inclusion of diverse cultures races and minorities to develop an unbiased and Equitable AI also AI company should ensure that they give due recognition to content creators

and there should be strict controls to prevent children from using AI unsupervised even now we have seen some examples where you know children AI had generated results which could have been harmful to Children next is openness and honesty AI companies if they start tagging their metadata for synthe IC content it can help reduce the misinformation Google has already ader to this practice and at a fundamental level users should have the right to know if they are interacting with AI meta got into recent uh got into trouble very recently with their open-source AI model llama 2 it responded to prompts with instructions to develop a biological weapon and now there is a debate to way pros and cons of Open

Source versus closed Source model noticeable data controls I think this is important all of us use social media and many other AI platforms but most of us forget to set our preferences for data security so that's where AI companies can step in and maybe they can make it more noticeable for us so that we you know when we are using this platforms very actively we can decide whether or not we want to share our data with them we can we can take the control of our data we can delete it whenever we want last is continuous vigilance so of course this is important and for AI companies especially having a defense in depth having controls at all

levels maybe a lot of of bug bounties pen testings internal controls uh external researchers taking uh their uh um expertise and telling them these all things are very important so last we move to us okay so the consumers I think many of us imagine a future like at L said like Sci-Fi movies and it can be a bit risky AI is integrating into our society way too rapidly our risk landscape is Shifting drastically too as it spreads to more it's as it spreads more within the smaller communities and among the less tavvy people the fact is that AI has been here for decades all the big sectors have been reaping the fruits of ai's potential it's just now that

individuals are harnessing the power of generative AI to enhance our productivity cut cost and more but generative AI is not Flawless we must pursue skills like fact checking data validation creativity critical thinking and a required understanding of the subject if we want to create high quality results so uh we have five minutes before uh um we end our presentation so I will just quickly go through all these points um first is education so like I explained that you know we cannot use AI we cannot rely completely on the AI there are two important things to notice here one is skill augmentation is required we are stepping into a world where we have to improve our skills we

have to upgrade we have to change our skills so this is one area where government can also step in spread the awareness about this as well as the risks of AI the second thing is AI should never be used as a front runner it should always be a

co-pilot also one more thing that I wanted to uh cover is it is very important to know what kind of AI you are interacting with so each AI company can have a different agenda it they can be profit based they can be sharing your data with third parties or they can be using your data for advert advertisements and marketing so make sure that you know your AI very well when you are uh interacting with it or when you are sharing your important data with it next is Vigilance so as we move more and more towards AI it is understandable that offensive threats will increase and with that our identity and data are very uh are at a

high risk so it is important to have regular cyber checkins also when you are sharing your data with AI like atad said you are sh you are putting it out there so be very mindful of what kind of sensitive information you are putting there okay so last is community empowerment so s like I said that the cyber security Market will change soon while we may encounter in increased cyber attacks there will also be a demand for more defensive roles only Community awareness can make the situation somehow manageable and we can play an important part in raising awareness so we can make sure that we are uh you know uh promoting an informed AI usage and we

uh we can foster a safer community so lastly to conclude this talk ai's expansion is inevitable and its implications are endless it's quite reasonable to say that AI will profoundly influence our future at this stage collaboration at various levels can be a Game Changer drawing from our own experiences we could not gge the depth of the topics that we covered today when we started the research for this anyways we hope that this session provided you some insights into this world so I think we are ready for your questions we have two minutes we all good yeah thank you thank you everyone yes right are any compies that are that's a very good question so I

recently heard that it is happening in the gaming industry so gaming industry uh has started using this um control that the children has to take a picture with the parent and only then they can use it it so you know this is I don't think any of the AI platform has started this yet but this can be implemented in this way thank you okay thank you [Applause] everyone