Data Privacy in the Age of AI Agents | Player1Taco - The Manifest Network

[Music] good all right you ready Taco all right Taco we're counting we are we are I apologize Eric how are you doing today I'm doing wonderful we're at the uh end of a really great e Denver here I'm sure I sound like I'm in a metal band I I think I sound like I'm in a metal band because I was smoking cigars all night uh for those that don't know my name is player one taco I am the CDO of the Manifest network uh and I'm here with Eric braik Eric hey nice to meet everyone so uh today we're talking about the importance of data privacy with AI agents um that's a hard topic like that is almost extinct because we give out permissions with everything we do what's sort of the overall thought of that yeah so if you look back at the Arc of History here it's probably good to start with what the fundamental problem here is uh if you look back through human history uh it took hundreds of thousands of years to uh organize basic tool sets uh for humans it took tens of thousand thousands of years for them to organize that to a level of standardization that you could get some basic uplift took thousands of years to organize into uh standards so you could get agrarian economies things like that took hundreds of years to organize into factories took tens of years to control all of factory output with software um and took now only years for AI to eat uh almost all of software uh for those of you who are in the industry I'm sure you know we've hit that inflection point of AI and AI agents uh where um you're you're now getting the vast majority of your ability to produce software through uh AI assisted means um so what each one of these epox shows us from a data perspective is that more and more data is generated um in each one of these uh at a geometrically uh increasing proportion uh and there is a further and further separation at each of those epochs between owner the concepts of ownership and control so uh you know there's a big difference between owning your data and controlling your data um and AI agents are going to accelerate this to an extreme degree so you talked about an interesting piece there of ownership and control right now a lot of people give up permissions to interact out of ease uh connecting apps and so they're giving up a lot of permissions for their data um what is that big difference between ownership and control yeah so we this is the difference between data ownership and data sovereignty is usually what we say instead of control um of the two of this is often misunderstood of the two of the sovereignty is far more important than ownership so we see this mistake uh in the set of manifest in incubators all the time we see probably uh five to 10 data privacy startups uh a week or so coming in to the platform um I would say 80 to 90% of them misunderstand that data ownership is really not the the Salient uh topic uh to solve for it's really all about control of the data and let me give you an example of that so um you may feel that you own the data coming off your AA ring for example probably you have one I have one probably half of the people here have one right um but ultimately that data is not in your control um you may feel that because you own it uh you should be able to derive some value from it which is what the premise of most data privacy platforms is is ownership equals value uh and that value should come back to the user but there's a fundamental misunderstanding here of the value is created by the control and the value is created by the product cycle that only the controller can move the data through so it's really the controller that generates the value and so the the value stream goes to the controller and that's probably correct in that case but there's so many platforms today that say that I get to control and own my data through tokenization uh and that AI platforms now allow me to gain value from my data is is my data individually actually valuable so there are some very good Platforms in this space I don't want to neg this space we should be doing this um and there are some very good examples one off the top of my head is is Vana um and I think they understand data and so they're going to be properly successful in that space um it it it your data individually for most use cases is actually not valuable so that's one of the common misunderstandings there's a game theory problem here which is is if I'm running if I'm a a curator of data and I'm building say large machine learning models AI models um or even doing traditional data analysis if I have a subset of users that want to opt out of that uh usually at scale that's totally not a problem for me like uh uh so from a game theory perspective I'll just take the users that want to opt in 99% of the time so from a game theory perspective the users that want to opt out don't matter they don't have any value to me so I'm not going to pay them um the second thing is that your data individually you may view it as valuable but it's uh not valuable from an analysis standpoint um as an individual under normal circumstances so I have plenty of data I do the analysis um and if you're not in the data set that's fine uh so again I'm not going to pay you so this is just simply a fundamental value proposition problem now that doesn't necessarily mean when I have say an AI model that I don't need my data right to put into the model to say where where do I fit right so if I'm doing like a health model for example um I need my data to figure out where I fit in there my data now becomes valuable to me but it's not necessarily valuable to the provider in aggregate the controller in aggregate so we're seeing a lot of different acronyms getting used about data privacy FH fully homomorphic encryption te's trusted execution environments are they actually actionable things right now in AI uh well that's that's a broader question so um there are actionable things that we can do in AI again I'm very Pro having platforms that try to promote data privacy but these problems are the problems of data are significantly complex so for for example um a lot of the data that is important about you is emitted by you but not owned by you this is this is uh metadata right so I can tell a lot about what's going on from the metadata that you emit that's not data that you own and it's data that someone else controls so this all of these Concepts really come down to sovereignty and and not ownership right so uh that's why we work in this sovereignty space with our partners um and really talk much more about sovereignty when we're talking about data do you control those emissions right how much of them do you you control uh do you control where they go um is that even possible right now keep in mind we're entering a massive accelerated uh stage of growth right now we're if you know where we are on the curve of AI acceleration right like like if you think we're at the inflection point that was a year ago that was 18 months ago for for most of those things we're way up on that curve now so the the more data is being generated at a geometric Pace than ever before you know geometrically accelerating pace so um and most of this data is not data you could ever claim ownership over but you might be able to get some control of some of it right so when we we're talking about the control ownership and sovereignty of data we have different protocols out there uh that provide Edge compute you know people at home can provide compute for inference for AI agents uh but manifest comes at at a little different angle of like Enterprise grade data center quality compute is that setting a precedent for better control and ownership of data well if done correctly so you this is this has actually been an an an issue with most things that we've tried to do in in blockchain is um if you aren't starting at fundamental layers of Hardware if you aren't starting at l0 you've got a lot of really significant uh data privacy and sovereignty problems if you're just implementing from the top of the stack down um there's a lot of promising work going on in ZK and other Technologies there's a lot of problem uh work going on with uh various types of trusted Computing platforms um a lot of these are about access controls and encryption uh that doesn't necessarily directly translate to to data privacy in the way that you you think it was it would especially with AI agents so imagine a world where you know there's a thousand AI agents running every aspect of your life you have a super agent which is running all of those agents and those agents are collecting and calculating massive amounts of metadata about you um many times I actually don't need to know the data that you are attempting to hide from me through encryption and these methods I know everything I need to know by the metadata that you you accrete around you in your data accretion disc right if you think a lot kind of a black hole concept right even if I have perfect encryption of my data uh all of the products I use all of the agents I'm going to be using the agents that are going to be running my life my finances controlling all my assets controlling all my coms um uh you know I'm going to be able to correlate most of the data about you uh from that so uh what really becomes important is sovereignty in this case that I have that I have control of those AI agents and I have control of them whether or not information about me in any way leaks right or is calculated by other AI Agents from a metadata standpoint so uh these AI agents you got to understand they're they're going to be faster at everything and they're going to swarm anything that they want so if they if someone wants to know something about you in the future or make something up about you in the future future that's not true um swarming AI agents are going to be able to do that across the entire spectrum of information production so um it's it's more about sovereignty over the super agent that runs the agents that run your life because the information is going going to be very difficult to uh to protect so if you're worried about protecting your data versus versus sovereignty worry about sovereignty first right so that's a little bit of a a bleer perspective uh versus what like Everyone likes to talk about like where AI is going to make everyone's life better in some way shape or form but then there's really cool projects like Venice or Morpheus that work to like give a little bit more ownership and control into people's hands of their data that they're using AI for now you're talking about a super agent or what we've talked you know named you know personal AI agent in some way shape or form what is something that someone can do today to sort of protect themselves or prepare well so let's start with I don't think that's actually o Bleak perspective because uh again perspective right because again if you're I'm going to combine the the two of these two things into one answer uh the thing that you can do right now is get really familiar with agent Frameworks get really familiar with running agents that operate on your behalf if you aren't already doing this you're already behind the curve right so get into AI get into AI agents um get on the Leading Edge of how this is working right and and get get comfortable with AI and AI agents running parts of your life because this is going to happen whether you want it to to or not right um make sure those AI agents are sovereign that that is the key um there will be many cases in which you will not be able to control your data and you will not be able to control the meta data that will reveal your data whether you encrypt it or hide it or not uh that problem is going to get worse so I'm not here to tell you it's going to get better it's not it's going to get worse but what you can do about it is grab AI agent platforms you learn to use them make sure they're Sovereign platforms so that you control the AI and then start getting into the zone of establishing control right forget about ownership forget about derivative rights forget about making Revenue off of that you know not that those aren't noble goals right and there's lots of great plat forms again as I already said that are trying to do that um but practically speaking right now you that's not going to help you that's a longtail thing what's coming for you really really fast is AI and AI agents that are going to run everything about your life around you um and that's going to happen whether you like it or not so you have to just be prepared for that reality um and I don't think that's Bleak because this is also incredibly empowering if you do this correctly your health will be better than it's ever been your mental status will be better than it's ever been your wealth will be better than it's ever been if you do it correctly if you so you're empowered this is an age of empowerment but if you do not do it correctly if you ignore it then putting your data in a data privacy platform by itself that's not going to do anything right right um and again not negging those platforms love that they exist partner with many of them they're great but it it's not going to help as an individual item um so before we wrap up we want to just shout out to eth Denver having us here to talk about this today sort of covering this Hot Topic as the first Speaker of the day today um before we wrap up with last thoughts here one of the last things that um we ran through too quickly in my opinion who are you and what's your history right so um I'm a neuroscientist uh I started about 35 years ago in AI uh after uh being in Neuroscience for a bit um then spent a lot of my career building Sovereign infrastructure systems throughout the world um so when uh when I saw Bitcoin in uh January 2010 became one of the early Bitcoin miners uh I really viewed blockchain and Bitcoin and Associated uh platforms as large scale behavior modification devices um I never really paid attention to the artificial money part I know that's what everybody's kind of really into but um I look them as large scale behavior modification devices so uh protocols so um for me I've been on a 14-year Journey since then of building fundamental blockchain infrastructure uh as an extension of getting to Sovereign AI right so for me my whole life has been AI in Neuroscience uh so I've always wanted true Sovereign AI platforms for people and here we are at the day that many of us oldtimers in AI have been waiting for where we're actually escape velocity right for a lot of this technology so um we're finally here I'm excited that we're finally here I'm actually really optimistic for the future um but I'm also a hardcore realist I've been through all these Cycles many many times and this does not turn out the way you think it's going to turn out uh most of the time so get get ready for not knowing how it's going to turn out as well rethink what you you think time is rethink what you think acceleration is rethink what you think data is rethink what you think ownership and sovereignty are um this is an age to rethink about the fundamentals of of how your lives have worked related to data up to this point because everything is going to change we are entering a new Epoch and at each one of those Epoch changes all the old rulers disappeared and all and a new class emerged and that's what's happening now awesome thank you and for those who don't know for some reason my name is player one taco and I talk about what other people do you can talk to Eric over there if you have any further questions thank you everyone