IBM’s Quantum Push, Project Arc and the Extension Surveillance Problem

Hey everybody, welcome to the Techstrong gang. We're going to talk about a couple of conferences today, specifically IBM and ServiceNow, and then we're going to have a little chat about, well, some latest and greatest security moves by folks that are going to surprise you as to what the bad guys are up to next. But anyway, let me introduce our guests for the day.

JP Morgenthal, good to see you as always, my friend. Hello there. And then we have Terri Robinson, who you can tell by the brick wall is back in New York.

Hi. Yes. Yep.

And John Schwartz, who I believe you're in Las Vegas for the ServiceNow conference, right? I am in Las Vegas. I'm looking at a major construction of a, looks like a Hard Rock Cafe.

It's a guitar being built to the stratosphere. Yes. All right.

Well- So by the room, okay, this is sad. Let's see if I get this right. You did not tell me- You can, I bet you- ...

you were staying, I'm going to say the room looks like the Venetian. Yeah. Which tower?

Oh. You have three options. The main tower.

I don't- By the lobby. Yeah. South tower.

The one by the food court. You're absolutely right. All right.

It's getting sad. Yeah. It's getting sad that people can identify hotel rooms- Right ...

when you're spending far too much time in Las Vegas. It's a very unique style. I love that room.

I love the- I can see it ... it dips down, you get a nice living room. You can work in there.

And it's unique to the- Right ... Venetian. All right.

But on the other end- It's very scary that I have done spoken at so many conferences that I know the room. That is scary. It's time to spend more time at home, I think that that's what the world is telling me.

Yeah, well, COVID did that. This was all pre-COVID. But I think we've all lost count of the number of times we've all been to Las Vegas.

Oh, God. So that's just troublesome in its own right. But on the other end of the country, in Boston, there was another conference, and this one was hosted by IBM, and it was its annual IBM Think conference.

And it was interesting on a couple of points. To be honest, IBM was kind of moving their AI ball slightly forward. They were showing some private previews and things that they previously announced were now in public preview, and it was kind of one of those little marketing initiatives that you see from time to time.

But what was really interesting was the CEO of IBM stood up and told everybody that this is the year where we will have real quantum advantage in the sense that quantum computers will be able to do things that classical computers cannot do. And this was the year to start building software, and he started pointing to some interesting things that the folks at the Cleveland Clinic are doing in terms of building protein molecules, or at least simulating them, which hopefully will lead to some more advanced research in the areas of healthcare. But JP, as you look at all this, have we reached a point now where quantum computing is a real thing that we can start building some software for and some applications, and is this going to be the next big thing after AI?

I think that it's at a point where it's functional for experimental use beyond just working and building and creating the lab, right? Like you said, they're doing some healthcare training of models using quantum computing. I think that's an interesting approach.

I'm starting to see that obviously, the scale of and the bandwidth at which quantum can operate versus even a GPU, is significantly higher, right? So everyone's like, "Oh, GPU and ASICs," right? That seems to be our approach.

I think it's interesting that quantum's now coming along, and the question for the ASIC vendors are going to be, is this a long-term strategy? And perhaps it isn't from a cost perspective. I think quantum is still going to be out of reach for majority of the world from a cost perspective.

I think you're still looking at it almost from a leasing perspective, which makes sense. IBM's coming at it, right? It's like their next mainframe.

You can rent and lease hours on our quantum computers, right? It's not like I'm going down to Costco and picking up the latest Acer quantum computer. So, I think it's on the journey.

I think it's good that they're able to demonstrate how quantum will be able to train specialized models and perhaps even be able to do that faster with lower memory footprint overhead. That would be fantastic. But I think that it's just in its inception, and I do think that there is a role to be played for other types of processors that will fulfill the AI requirements prior to quantum being generally available.

I think you're spot on, and if you look at it, they have three models, right? They're basically saying you can invoke this as a cloud service through some sort of API capability, or you can book time on our quantum computer. And IBM is busily digging a massive hole in the ground in Poughkeepsie, New York, to build one of the next generation quantum systems.

They claim they have 80 of these so far, but the next one's going to be even bigger and better. But there are reference architectures out there that they've shared if you care to go build your own, but you just need a few billion dollars in your pocket to go do that. So to JP's point, I don't think that's going to happen all that frequently.

But John, I'm going to throw this your way. One of the more interesting things that they did say was that the output from the quantum computer would be used to train our AI systems and therefore make the AI systems- Yeah ... relevant and interesting.

Yeah. Yeah. With all this chatter about quantum, we talked about the concept and potential uses.

This almost looks or sounds like a convergence of quantum and AI in a little bit. So we have, and correct me if I'm wrong, these quantum outputs to train AI models, which in a sense to me suggests this kind of hybrid future where the classical GPUs would handle the heavy lifting, and then the quantum units would solve these high complexity variables specifically. Does that sound right in a sense?

If that is true, with the orchestration becomes the control plane, who owns it? Is it the cloud providers or the AI model providers or IBM? All right.

I think it all depends on where it was executed and how. So I think you're kind of getting to the nub of how it is we're going to pay for all this stuff. But I would just throw this out here, and this is the part that I find a little, I don't want to say disturbing, but it's definitely making me raise an eyebrow.

So if I understand this correctly, we have a whole bunch of generative AI systems that we're not really sure how it works, right? We can't empirically prove how the gen AI generates its output. And then I look at quantum and, well, we don't really understand how that works either.

So we're going to have a thing that we don't understand how it works, feeding into something else that we don't understand how it works, and then we're going to go and actually execute on this stuff because, well, it appears to be the right output based on- Hold on, Mike ... Why are you saying you can't empirically see how the AI engines are working? Or I don't want to quote, you said basically you can't empirically understand how the LLMs are working, which- Yeah ...

I don't think is the case anymore, right? I think we're pretty near. I think we just had this conversation earlier this week with the professor of AI from Carnegie Mellon University, and this was part of the issue.

They were saying that they cannot still empirically prove how that LLM works in there. There's a whole field called interpretability that's emerged to go investigate this stuff, but they have not quite come up with the, here's how from point A to point C, how this thing reasons its way through this to come out with this particular output. So that's where I'm kind of getting that piece of that from.

And then I also see with quantum, we are messing around with a computer framework that essentially mirrors nature, but we don't really understand how nature works either. " What do you think, JP? Quantum freaks me out a little.

That's just the nature of everything I've ever been taught about computing is binary zeros and ones, and that goes away with quantum, right? You're now on a what, base-- I'm not even sure what the base is. Is it base 10 that quantum operates at, right?

So you're operating at the ability to represent information at such volume, in such a unique manner, and be able to do mathematical operations on that. I actually think that there's a good correlation between quantum and the discovery architectures, the encoder, decoder of the LLMs. Because it's so heavily based on looking at vectors and doing matrix multiplication, which quantum can do at an extremely faster pace or rate than even GPUs.

So, I think we can get a significant boost in performance, but the thing is, what's in your vector space is what's in your vector space. All right? Dog is in your vector space, that's it.

You got dog. Doesn't matter if the quantum process is doing it or the GPU or the CPU. It doesn't matter.

All I have is dog. So it's not like I'm now got giraffe all of a sudden because I went quantum. So it's garbage in, garbage out.

You can just do it fast. Yeah. So, Mike, can I-- Go ahead.

Well, I was just going to ask you, what is your hype spidey sense telling you about quantum at the moment? Because on the one hand, I got folks who are saying we're on the cusp of a golden era of IT, and it's going to be all kinds of interesting new solutions. And yet, I've kind of heard that before and multiple times going back- Yeah ...

to client server, if I really wanted to go all the way back in time. What you just said was something that Bill McDermott at ServiceNow, the CEO, said yesterday. He talked about this golden age of innovation in terms of IT and enterprise.

So when I hear that, my spidey sense, like yours, goes off. And here's the thing, and we're going to talk about ServiceNow in the next segment, but we have this whole kind of raft of these AI agents coming along. So as companies start deploying all these specialized AI agents, for me, I think about this concept of agent sprawl, where these systems maybe conflict or bypass security protocols.

And then I think you layer that with quantum and some of the unpredictability, and it raises issues for me and makes me a little bit paranoid. But then again, I am a paranoid person, so that's kind of where I'm coming from. Speaking of paranoia, Terry, the security people are watching this quantum stuff and are probably wondering, is this Q-Day thing going to be real?

But one of the things that occurs to me as I look at all this stuff is, we've talked about this in previous episodes, but Q-Day was supposed to be in, I don't know, 2029. So it's tomorrow, Mike. It's tomorrow.

Yeah. Well, that's what I'm asking, because it feels like the advances in quantum are moving a lot faster than anybody completely understands. Yeah.

I'm sure our friends in China are after this. Yeah, and it does seem like, I just poking around in the last week on this topic, it seems like people have shifted, and there seems to be an expected acceleration, right? So, I don't know.

So then we're turning to quantum to solve the problem that quantum's going to create with cryptography, right? So- That's a classic IT business model, yes. Yeah.

So yes, exactly. So I don't know. Maybe the speed is going to help us there.

But also, I'm all excited about quantum on one sense, but I have the same reservations as you guys in it, particularly when it comes to security stuff. Think how fast quantum moves, and the fact that you could find dependencies really quickly, and whatever. But that intersection between quantum and AI is troublesome.

And I just think it's moving so quickly. I was looking at some... There are some advocates out there, like do you guys talk to Tom Patterson over at Accenture?

Because he's been really advocating for quantum for a while. I'd love to maybe get some of his opinions on the security side of things. But they seem to feel like this is just coming on.

I don't think we're going to be waiting till 2029. We'll see. So JP, here's the thing that I'm also trying to figure out, and I'd love your opinion on it.

So we're waiting on this AI side, we're waiting on AGI. We're going to basically come up with these smarter reasoning AI agents, and that's going to be great. And it might take us another few years to build out the infrastructure to actually, maybe, possibly achieve that goal.

But will a lot of those efforts essentially be obviated by the rise of quantum because some of the things that quantum can do are some of the same things that we're hoping AGI will do? Or do these things, are they complementary or is there overlap there? Well, for me, AGI is different technology than generative technology, right?

Parts of it will leverage the neural net technology and capabilities, but I think we've proven that AGI is... I'm sorry, that generative is not the basis for AGI. Will we get there?

Will we be able to... So AGI is really based on being able to reason. And in order to reason, we need to be able to have semantic and contextual understanding.

I think with generative, we get contextual, we don't really get... We emulate or simulate understanding value. So no, I don't see it driving AGI right now.

I think we first need to solve the problem of knowledge and being able to represent knowledge in order for AGI to truly be able to reason and understand. So semantics and being able to represent semantics as part of the information base that is available and natively processed is key. And I don't think we really have that.

Well, is the industry looking at AGI and quantum in two different ways? Like AGI to... Well, who knows?

But I saw that shoulder shrug. But AGI, there's a lot been put on that about it sort of replacing human agency, right? But that's not how quantum is thought of.

Am I correct in that or? I think you're correct in that, but we don't know what quantum can do, and we're, again, not even sure how it does it. But- But who's doing it?

at some point, somebody's going to try to reason, I don't know, theoretically, I'm just making this up, but ask an AI agent to reason through why there might be a cancer cluster in a particular region and what in the environment contributes to that, and what from our DNA contributes to that. But theoretically, the quantum machine could do the same thing, right? Maybe do it simpler or maybe do it in a way that replicates what's actually occurring in that cancer cluster.

Again, I don't know if anybody's working on that or what the theory of it is. You think? It just seems to me that there's going to be some overlap here, and we're going to have to sort out what it is we're doing.

And are we going to spend a boatload of money on AI one morning and wake up to quantum has basically figured it out already? Well, so let me throw this in. The thing about quantum that I keep seeing is the speed at which it can process.

It's like quanta FLOPS or something like that. So it's been very focused, in my opinionOn the data that I've read, in articles that I've read on speed. For example, potentially, a quantum computer could break the blockchain encryption in as little as nine minutes.

If you try to compute that against what today's GPUs could do, it's years. So you're looking at an unprecedented scale for extremely complex mathematical operations that can be performed. But I don't see how that translates to now opening up the universe of AI unless we have a mathematical limitation that I'm not aware of holding us back from reaching AGI.

And possibly we do. I'm not a researcher in that field. All right.

I think we firmly established that we don't know what's going to happen, but- And we don't know who's going to be doing it. We're way behind. There's just too many variables at this point.

Right. But I only had one question, though, maybe JP, you could answer. What is the timeline- I'll try ...

for AGI? What are people saying about it? Oh, we've heard everything from next year to 2035.

Okay. I'd be more concerned in your area, Terry, what I'd be looking at is you couple Mythos with Quantum, and- Mm. Yeah ...

I'd run for the... I'd start looking to- Yeah ... colonize Mars.

And that's next. They can start the world very fast. You go up there and set it up.

I hear Elon has some openings in his- Yeah, exactly ... company. Let's get those SpaceX shuttles ready, baby.

Well, that also just raises, and there's no answer to this, but we've said this with AI and everything else that moves quickly, especially AI. With Quantum, what are the guardrails there and is anybody working on imposing them in any way, shape, or form, or is this just going to be a free for all? Yes.

Yeah. Isn't that Spyglass or whatever, the consortium of vendors that were going to at least take a look at Mythos and make sure that anything potentially zero day in their own software was not at risk. Right?

And then, there's been no release schedule for Mythos as a model right now. All right. I think we can- No, they probably- ...

firmly say for sure, though, that three amazing things are happening. Well, two, depending on how you want to count, but the three amazing things are Gen AI, AGI, and Quantum. And instead of focusing on just one of those things, maybe we should be paying attention to all three of them because that's what's going to drive computing for the next decade.

But I think sometimes we all get obsessed with one particular flavor and not look at everything else in its totality. So that's my two cents on it. I'm going to shift a gear here because John is at the ServiceNow conference in Las Vegas, and we typically like to get a little report from these events when we go visit them.

So John, I know surprisingly that there was a conference, and NVIDIA showed up, so that was shocking in its own right. But what- Of course they did. Yeah.

So it's an annual event. Yes. It's an annual event where Jensen Huang is a surprise guest and everyone is shocked, except the people who attend.

So what- ... ServiceNow did and what they do every year is this kitchen sink series of announcements. And so I rooted through all of it and I'll lead with NVIDIA because you mentioned it, Mike.

There was a key announcement which was the Project Ark, which is an agent that resides on employee workstations. It's built to execute complex multi-step tasks independently. So because there would be security concerns obviously about this in giving AI the keys to a desktop, the agent's secured by NVIDIA's Open Shell runtime, and it's managed by ServiceNow's AI Control Tower, which is a centralized command center for every AI system agent and workflow within a company.

To that end, ServiceNow also made some upgrades to the control center, the control tower, which include this literal kill switch capability. So they're using their security company, Veza, to monitor AI behavior in real time. So if the agent goes off script, so to speak, or attempts to access unauthorized data, they can detect this and shut down the process allegedly instantly.

So there's that. The thing that I found more interesting was later after the keynotes, Bill McDermott did a fireside chat, and he talked about a couple of things that I thought were interesting. He mentioned some stats.

One of them that jumped out at me is that only they find among their enterprise customers and enterprise customers in general, only six in 10 have a strategic agentic motion. They've started an agentic motion strategy, yet only one in 10 have any tangible results so far. He also mentioned that, and I found this really interesting, that half of their revenue comes from consumption-based pricing, and that he mentioned more than a billion dollars comes from security.

And he predicted that by the year 2031 or even 2032, most of their revenue is going to be through security-related services. So he's almost trying to position the company as the AI security solution, so to speak. Again, we talked about this in the green room earlier, but this is a company that pivots conveniently to where the market is headedWith a fair amount of marketing zeal.

But again, it's an interesting show. Jensen Huang was here again, and every time he's here, the joke is that every time he appears here, NVIDIA has gained another trillion dollars in market value. So he refers to Bill McDermott as his trillion-dollar lucky charm.

All right. JP, what is your sense of what's going on with these guys? Because when I look at them, I go, at the heart of this thing is an IT service management platform that's based on a proprietary architecture that they have created that then allows them to extend that into different processes and workflows so you can get an HR app or you can get a CRM or whatever else it's going to be.

And on the face of it, that allows you to reduce costs because, well, you don't have to go and build separate stacks for each of these applications, and that sounds reasonable on the face of it. And yet, we still have all these other applications that people built for different things and different... Frankly, there's just a lot of sprawl out there, and for whatever reason, the world seems to resist this move to centralization.

So why and-- Or is this just an ITSM platform that's trying to be bigger than it should be? I think it's an attempt at survival, to be honest with you. Yeah.

I know for a fact that one of the number one applications under attack by AI is ITSM. Because let's face it, Claude does a lot. I've used Claude in building as my-- It's managed my entire DevOps process.

It's done the CI/CD. It's done the provisioning of the hardware. It'll set up operations and monitoring.

It has access to skills, and that can touch things like Datadog or Splunk, and integrate with those apps out of the box now. So, everything I wanted to or before used to use ServiceNow for, if I were a customer considering a move based on the cost, the subscription of what ServiceNow was in the past, I'd be like, "Hmm, can't we do that with Claude? " And the answer is a lot of it, especially now with the computer use stuff that's built in.

It's incredible the amount of capability and flexibility that you have to automate various OpenClau, Nemo Clau, right? We just talked about the NVIDIA stuff. There is incredible number of options on the table.

" And so I think that's what he's doing. He's acquiring, he's putting pieces together, and he's trying to build a moat to fend off AI and make sure that his existing customer base doesn't start to bleed off into the LLM world. Mm-hmm.

And to your point about that, and they're not the only ones guilty of this, John, but it feels to me like there's a lot of vendors out there who are trying to answer the relevancy question that JP put- Yeah ... where basically, they have some sort of legacy platform, and now they're trying to position that- Right ... " But will all that stuff ultimately just get replaced by something that is truly AI native or will they re-engineer these things to still be relevant?

So there is like a level of resistance and there's an undercurrent. And one of the interesting thing during this 45-minute Q&A with the reporters was one reporter in particular pressed him around that issue, and the normally unflappable McDermott was somewhat taken aback by this challenge from him. And it was an interesting insight because usually he doesn't lose his cool, but he was in sell mode to kind of explain the relevance.

And I think about this, and it's kind of alluding to what JP was talking about, but I'm starting to think that-- ServiceNow talked about how their agents work across AWS, Azure, SAP, Oracle, and I always wonder, to me, it always comes back to can a single control tower truly manage security across a fragmented ecosystem? Or are they just creating a single point of failure? And I'm sure the enterprises are thinking the same thing.

So the irony is he's talking about security, but yet there's also some underlying issues. And I really do think ServiceNow, given their history, and it's not just McDermott. This goes back to John Donahoe.

They have a history of trying to kind of swim or move to where the market goes. I'm not saying they're going to fail or I'm not saying they're going to hemorrhage customers, but I think this is part of their DNA. I don't know.

Terry, can there be one rule to rule them all? No. I don't think so.

But the short answer, I think it is interesting because I think you're right, John, that ServiceNow moves To where the market is, and that's how they stay relevant, right? And they've had some success in their past ventures of doing this. I am among those that have been concerned about the idea of a single point of just sort of failure.

But I also wonder in cases like this, if they're saying all the right words about sort of guardrails and kill switches and things like that, if that's going to appeal to their existing customer base, and if maybe sort of reputation alone or relationships alone are going to win the day for them, maybe not in the kind of numbers that they want, but- Oh ... is that going to- I think a lot of it will also come down, well, there's two things in my mind that are at play here. " And I think if you ask IT people, they're like, "Yeah, we want the platform so we can have one single place to manage all this stuff," and that appeals to them.

But when you go talk to the end users, they're not always as enthusiastic about all these different- They want best of breed ... Yeah, because they want best of breed, and the HR people want an HR app- Yeah ... designed by people who know HR, and that may not be the case when I get something that was designed for IT people per se.

So- Well, yeah. That's an age-old conflict in tech so. Well, and I don't think their messages resonate...

I know it's not resonating with Wall Street, for what that's worth. The stock is down like 30%. Yeah, which comes to my next point, because I think it's really related to this as well.

We've talked about this, and I have trouble saying this word, SaaS apocalypse. SaaS Apo- Apocalypse. Apocalypse, yes.

Yeah, apocalypse. That's it. Apocalypse.

SaaS apocalypse. And that issue is what JP was talking about, which is, in the future, am I going to need to have this kind of app that was designed for humans with this GUI, or am I just going to go to an AI agent and tell it to go perform a task, and it will invoke the stuff from the back end, and it will complete that task? And I know that all the people who build SaaS apps are trying to build walls around their APIs now to make sure that an AI agent can't just willy-nilly do whatever.

But JP, as you look at this going forward, it seems like to me, the AI agents are just going to do what we tell them, and we're not going to be concerned about where exactly and what it was that they invoked to do it. You mean we're getting towards a level of autonomy where- Yeah. Mm-hmm ...

they're more mission and task-oriented than they are required execution of a detailed statement. Right. Agents don't care about what the application experience was.

No, and others have spoken about that as well, right? Look at what's happening with the web. I think that was the first one to be attacked.

Like all this web advertising model is now threatened by the fact that I have headless web browsers going out there and sucking data off pages, and then returning it through a chat interface. So this is just an extension of that, where I have that ability, where the LLMs are doing the work. It really doesn't matter to them whether the experience is good or not, as long as they can figure out, as long as it's intuitive.

Almost gets back to why am I going to waste so much time on making it look pretty? 0, where it's easier for the other apps to read. And maybe even I create a bot interface, right?

And we direct the bots if they're doing the work on the webpage, automating that step that, "Hey, use this link, not that link for you. " Well, folks, I'm going to leave this here because, we don't know how this drama's going to turn out just yet, but there's going to be drama. There's going to be some agita, and we're going to be talking about this, I think, for at least another year until we see how it all gets sorted out.

But I am going to shift the gear to our next topic because Terry Robinson has an article on Security Boulevard talking about how extensions to browsers are being compromised to create these kind of, well, I don't know what else to call it, but it kind of looks like to me almost virtual botnets that I'm sticking in my browser, and people are taking over my systems in new and interesting ways. Terry, what's going on? Yeah.

So, the browser is where all the action is these days, and it's going to be the hotbed of security according to a lot of the people that I've spoken with. But in this particular case, browser extensions are going through, and they're collecting and reselling user data, right? And this is what we would have called back in the day, not so long ago, spyware, essentially.

But the thing is, it's perfectly legal, right? They have permission to go in. Sometimes, users don't even know that they've given their permission.

Maybe they have given permission to some other entity, which has now passed that on to something else. And this is all from LayerX research. They found networks of these extensions, like they say more than 80, but I'm sure they can find well more than that.

Some of them were media extensions. Like if you have Amazon Prime or Disney+, and those kind of things, they're all there legitimately, but they go outGet data. They fill in any gaps in information.

If the user doesn't give up certain data, then the extension developers can match email addresses against some of the third-party databases that have demographics. And then if you've ever been hacked or your credentials lifted, which has happened to all of us, they can put together a pretty good picture of who you are, and these things just spin off on their own. So there we have it.

You're being spied on by Disney extensions. So, I don't know. JP, I'll throw this your way.

Can I kill all my extensions? And as you look at all this, what does this mean? Or is this just a new fact of life and we're going to have to suck it up, buttercup?

No, I actually think that there requires some policy that needs to be created about the use of this. I think there's a lawsuit probably coming when people realize it. And I get them all the time.

I got one today in my Facebook stream. com may have stolen your information. So the lawyers are out there, they're hunting, they're looking for you.

This one stole your information, that one stole... And not only that, but there are settlements. I've gotten quite a few little drops in my PayPal account thanks to some of the...

50 from the settlement. I mean- There are consequences. It's known, it's resulting, but the punishments aren't severe enough that it's worth giving up.

So I think it's a machine, and it's feeding itself. And the penalties- Yeah ... are just part of the system.

And so we all benefit in the end, right? Hey, you stole my data, but I'm going to join a class action suit, and I'll get paid for it. That's great.

You should pay me in the first place. Well, but these extensions are doing it legally, and then it's being amplified by AI, which is going to lead to SkyNet. Legally is questionable.

Again, if it was legal, they wouldn't be losing or settling the case. The problem is that they really don't want the bad press, because regardless of- Right ... whether they did it legally or not, it doesn't look good to the public.

" Just- Mm-hmm ... not what Josh Tomorrow needs right now. It's just not happening.

So sure, $30 million? Yeah. Here.

I'll put it in an account and share it. Wait. I want to sign up for a new web service, and here's how it's going to work.

So basically, I will sign up, and your web service will scan my browser to determine what extensions that I have, and then you're going to scan whatever lawsuits that are out there related to my extensions. And it will automatically sign me up, and then it will just send me a check, and I'll never have to do squat. What do you say?

All right, Mike. So you come up with these ideas all the time. Let's pursue that one.

Is that a great idea? That actually seems like it has some legs there, and could take off. So you think of this as a consumer problem, which it's also a problem for companies.

Because they consist of consumers as their employees, and they may or may not have some of these... There are other media things that they might legitimately have extensions for on their work stuff, but a lot of them have Disney Plus and Amazon and Prime and all of that. So it is kind of, I think, JP, you maybe touched on it, to set some policy and also to have some teeth, not just in the legal sense, but within an organization.

Some of the people I talked to said there's the basic know what extensions exist, because most people don't, and a lot of times, IT doesn't look at that necessarily- Mm ... in this day and age. And look at the policies of these people that own these extensions and say, what do their policies say about what data they own and they can collect?

You need to do that. So I know we're kind of coming up against it, but can I ask you a quick question, Terry? So given this discussion, should enterprises treat browser extensions like endpoints instead of user preference tooling, or?

Maybe so. That might be the- Maybe ... way to go.

And then that gets back to know the extensions that you have, treat them as such. Yeah. And to look at how the, and I guess you would do this with an endpoint, look at how those browser extensions fit into workflows.

Right. I would just say that anybody who's offering you a quote, unquote, "extension" is somebody who built something for a particular reason, and if you're not paying for it, they probably have an ulterior motive. And also remember that extensions have extensions, and it can just spiral out of control.

So, go take a look at your machines and decide what's real. Or sometimes, what I wind up doing is every couple of years, you get a new machine and you start over again, and that's probably the most clear moment I've ever had. " Exactly.

Know what it is you're accepting. Folks, we got to go. I want to thank our guests for sharing their knowledge and insights.

As always, we live in the most fascinating and interesting times of IT ever, and you just got to be a little careful out there is all I can tell you. But hey, stay tuned for the rest of the replay of the Techstrong TV lineup. It's going to be awesome, just as it is almost every day, and we'll see you all again tomorrow.