Microsoft Security Copilot Blocks AI-Generated Phishing Attack | Tech Field Day News Rundown October 1, 2025

Check out open ai simple zero day exploits. Prime refunds. HSBC goes.

Quantum aura is down with MCP security. Copilot doesn't like phishing, fair waters in Wisconsin. And Agent WebEx is here in this episode of The Rundown.

Hello everyone and welcome to the Tech Field Day rundown. Congratulations. It's October we made it, and we are here with Watts of great news for everyone out there who's very, very, very excited to hear about AI and breaches and maybe a couple of other things.

I am your host Tom Hollingsworth, and I hope that you're ready because I think today is pumpkin everything day. It is pumpkin seed day, pumpkin spice day, even though we technically start pumpkin Spice Day in like August Now, um, it's also random acts of Kindness Day and my random act of kindness is to bring you my amazing cohost, Mr. Alistair Cook.

Al, welcome. Thanks Tom. And I'm not quite sure who you're being kind to by bringing me here.

Uh, on the other hand, it is also national homemade cookies day and maybe don't pumpkin spice all of the cookies. Yeah. And uh, one thing that does definitely go well, that doesn't have pumpkin spice in it is National Jiffy Mix Day, and in the Immortal words of Chris Rock Cornbread, ain't nothing wrong with that.

There ain't nothing wrong with our news stories this week because we have quite the lineup. There's been some exciting things that have been going on in the industry and we definitely wanna bring you all of the fun that has been going on. But first, we're hoping that you're gonna buy something, or at least OpenAI is because they have added new instant checkout to their US users, and that allows you to buy directly from currently Etsy, but soon Shopify.

And, uh, you could do this without even leaving the chat. Purchases can be made with Apple Pay, Google Pay, stripe, or your trust old credit card. OpenAI, of course, says that all of the results are organic and that the merchants are only paying small fees while sellers are still gonna handle orders and payments.

It's also going to open source, its Agentic Commerce Protocol, a CP, to help spread adoption of buying things from a chat bot. Uh, al they're hoping that this is gonna put it in competition with Google's very similar system, but are you ready to trust chat GPT not to order 14 cases of cat litter? No, I think there's nothing wrong with having 14 cases of cat litter, uh, particularly if you've got a cat less, less useful if you only have children.

Uh, that's the interesting part, isn't it? I, you know, we, we have limited faith in the capabilities of, of chatbots and their ability to not hallucinate and make stuff up. And sometimes they'll apologize for making stuff up, but, but still doesn't stop them making stuff up.

And as you say, ordering vast amounts of, uh, of product might be an, an interesting, uh, mistake to make. And who's really liable then if that 14, uh, crates of, uh, cat litter turns up at your front door that you really thought you only ordered one. And I wonder if this is more of an attack on Amazon.

'cause then chat GPT becomes the one storefront for everything that Amazon wants to be. Uh, it really will be interesting to see if this rolls out to wider and wider locations if more and more companies end up sitting their marketplaces behind chat, GPT, uh, 'cause I can't say Amazon wanting to hand, uh, open AI a bit of money on, on every transaction that they're taking a bit of money off every transaction from their customers or their, uh, their merchants through. So, uh, I think that might be a little bit of a challenge along that road.

But in terms of having an interface to simply asking enriched chat type questions, and the example is like, what should I get my friend who loves ceramics as a question rather than, how do I do a text-based search for products that would be good for my friend James, who loves ceramics? Uh, you know, try to craft your search terms to get to the right products is sometimes a little challenging. As you go through the stores, you get millions of lines of things that aren't quite right.

Uh, so maybe chatbots are gonna be helpful here, but do I really want to be handing my credit card over to a chat bot, particularly if I'm on one of the free tiers with chat GPT where this is available, where they own every piece of data you hand to them. There might be some legal implications to them owning your credit card number, let alone the risk to you. Uh, it will be interesting to see, again, whether that's standard for the, um, the agent commerce protocol that we'll do this interaction between the, uh, open AI platform and the individual merchant platform.

Whether that gets adopted somewhere, or as we're seeing in most AI parts, 15 new standards will evolve a long time and, uh, then will end up with every standard under the sun, none of which quite work together. Uh, there's also a move from Google around AG Agentic PR payments protocol, a AP two they're calling it, which is for AI generated, uh, and initiated purchasing. So there, there is going to be a place in here.

There is probably a, a place where money is going to be made from AI by having it involved in all of your sales and purchase transactions. Those of you with a long history will recognize the acronym SE and MP as standing for securities. Not my problem.

It's come back again. An estimated potential 2 million Cisco devices are at risk from a newly discovered zero day floor where there's a, uh, a stack overflow in smmp, a network management protocol that's notoriously not very secure, and that any sensible person wouldn't be exposing to the internet. But many, many companies have, uh, Cisco, iOS and iOS XE devices are susceptible to this fault.

And Cisco has released patches and strongly urges customers to update immediately. Gotta think that the customers who expose this, their, uh, switches over the internet, uh, on snmp, are probably not the customers who are the most UpToDate with patches. Uh, this is also one of 14 floors fixed in Cisco's latest security update.

It seems like if you're good at it, you're not gonna be affected to by this. But lots of people aren't good at it, are they, Tom? Not if they're exposing SNMP to the internet.

I mean, if you look at S-N-M-P-V one, it, it doesn't even authenticate anything. Uh, at least in V two, we had some kind of a user authentication system to prevent people from doing well. Pretty much the thing that we're expecting this exploit to enable.

Um, if you're doing this, please stop. Like, like I get it. If you, if you don't have a better way to manage your, your information, then you have to use SNMP.

In fact, there's still a lot of things that just basically rely on it, but please don't expose it to the internet. There is no reason that anybody that jumps on showdown should be able to get a hold of your SNMP information. Oh, I know what you're thinking though.

I'll be sneaky. And instead of using public as the key string, I'll make it Well, whatever public written backwards is, yeah, guess what? That won't take me long to figure out.

This is an example of a bad configuration decision being compounded by an exploit in the stack of a piece of software that probably hasn't been updated for a very long time. Because SNMP is a solved problem. It, it works and we don't need to update it anymore until we find a one of these little bugs, okay?

So we can prevent a stack overflow, um, that if you read any textbook talking about how to properly exploit anything, you know, stack overflows are the first thing you go to, right? Let's, let's try to grab memory that doesn't exist. What can I get out of this?

What can happen? And the fact that stack overflowing SNMP, uh, process allows you to run commands is root. Who?

That, that, that's a rough one. So yes. Uh, in order get SNMP off the internet that will fix a lot of your problems for the future.

Uh, failing that patch, patch and patch some more. And then, um, really look at why you're using SNMP. Uh, there are probably better options out there right now.

Yes, some of them are probably gonna cost money, but trust me, you don't want to be paying for breach insurance premiums as opposed to, oh, look, we have telemetry now and this works so much better. So again, in the words of my good friend Bob Planker, patch, patch, patch and patch some more. And, uh, yeah, then security might actually be your problem.

5 billion to help settle some Federal Trade Commission claims that it might have misled customers into prime subscriptions and made cancellations very difficult. 5 billion in refunds. The FTC said that Amazon's checkout and multi-step iliad cancellation process tricked customers, including fun things like a button that says, no, I don't want free shipping.

5 billion, didn't have to admit to any wrongdoing. Millions of prime members are gonna get automatic refunds and others are gonna be able to file claims to get more of those refunds. This case, which dates all the way back to 2021 highlights growing regulatory scrutiny of Amazon's practices.

And I'm sure that if you wanted a copy of all of the legal paperwork, it could be delivered with next day shipping if you're willing to pay enough. Al is it getting harder to cancel things now? I definitely feel like it is.

It's not just the gym membership that's hard to get rid of when you paid for it and committed to keep paying for it and no longer use it. Uh, it's, uh, across the industry. I don't think this is an Amazon specific thing.

Uh, we recently went through a, a large number of hoops to ca cancel a meal delivery service because they really didn't want you to cancel. They made it easy for you to defer for a month and then be charged again. This kind of behavior is, um, pretty unpleasant to consumers.

It is all about retaining customers and retain that recurring revenue, even if you're not delivering any value for that re recurring revenue. So yeah, Amazon having a tortuous process to get off prime and to cancel your prime subscription, as well as having a, a very integrated process to get you to subscribe even if you didn't really think you would by telling you that, uh, your prime subscription is gonna be paid by paid for by the second time you're paying, not paying for delivery. Um, yeah, maybe that's, uh, a a little bit aggressive for them, particularly since the, the, the user interface was set up to tell you that it would be a bad thing to not go and commit to long term buying the, uh, the subscription that, uh, can be a a $15 monthly service or $140 annual fee, um, and just turns up.

And then you think to yourself, I must remember to cancel that. And you start the process, find it too hard and think, I'll get back to it later. Uh, the settlement was made once the legal proceedings began, but the legal proceedings then did not continue.

So Amazon cont continue to say that they've complied with the law at all times because of proven. Uh, the investigation began about four years ago in the, uh, previous Trump administration and, uh, continued onwards through changes of government. Um, this, this is a pretty significant thing for particularly the, uh, 30 million subscribers to Prime who will get their $51, uh, refund within the next 90 days.

That's the one and a half billion dollars of refunds. Uh, probably helps the FTC to get their billion dollars out as well. Of course, it's all a drop in the ocean when you're talking about single figure billions and Amazon and, uh, the AWS service as well.

So, will we see big changes in behavior? Mm, no. We'll see a change in the behavior that was specifically mentioned, and we'll continue to see Amazon being a large organization that wants to look after its shareholders.

HSBC has used IBM's quantum processor to improve bond trade predictions. Hmm. Getting better.

34% better at estimating whether a trade would execute at a quoted price. Hmm, definitely sounds like something computers should be doing now. Quantum computers are doing it.

Uh, the trial used mirror data from, uh, real market conditions, looked at the prediction but didn't actually place any trades. It's not yet been used in live trading. Uh, however, seeing a 34% boost in the, the accuracy of, uh, estimating whether the, uh, trade would close is probably a good thing.

Um, but complex financial, um, predictions and, uh, potential for improved financial returns from this investment in quantum, we might get a new quantum bubble to make the AI bubble looks more. I'm breaking some news right now. Al, I've, uh, consulted my quantum computer over here and I've uncovered a pattern, a very important pattern I wanted to let you know about it.

Uh, grifter ISS gonna grift. That's the pattern. Um, yes, I know that the next thing after AI is going to be quantum.

Why do I know this? Because eventually all hot markets cool off, all bubbles eventually deflate somehow, and then the people who want you to buy a thing need you to buy the next thing. And it sounds to me like quantum is where that's going to land.

So let me give you a little example of why this story does not really pass muster with me. One, the reason why quantum computers are so good at what they do is because they don't think linearly through problems. So, for example, instant factoring of cryptography is because they can factor all prime numbers instantaneously, which means they can get the secret keys.

The problem is that those are static inputs, and when you start monkeying around with the inputs and introducing uncertainty into what's going on, quantum computers really can't keep up with that. Um, you know, one of the ways that we've made encryption algorithms quantum resistant is by using lattice, uh, encryption algorithms where you get to a decision point, you make a decision point, there are branches off that decision point, and I pick one of those and, and basically you're, you're collapsing all probabilities down. Uh, and quantum computers suck at that because there's, there's too many variables to analyze.

So did they get an awesome result by feeding a simulated bond trading market into a quantum computer and then having the quantum computer pick out certain values and, and seeing if they matched up with reality or in this case a simulated reality? Absolutely. Do you want to know what it couldn't simulate?

What happens when all of the trading algorithms in the actual real market find out what's going on and start adjusting the way that they make trades based on what this quantum bond trading program is doing? Because we've seen this before, right? Anytime you just go to Google and search for flash crash where something as stupid as a rumor that Steve Jobs had passed away causes the market to fall to a point where the circuit breakers kick in, why are those circuit breakers there?

Because most of the trading now is being done by computers and by algorithms in that case, what we're seeing is the race to find a better algorithm because that means that if I can execute the trades faster at certain prices, I can use the arbitrage in the market to make more money. And if all of that sounds like Greek to you, you are not the target for this. What they're hoping is, is that you're gonna buy in to using quantum computers to make a little bit more money a little bit faster than all the other Porsche schlubs out there.

Retail investors are really being left out in the cold. Most of the high powered get rich quick stuff is being executed by algorithms. And the hope now is that they can use quantum computing to accelerate those algorithms even further.

But that's not how quantum works. There are too many variables in this equation to balance it out quickly and cheaply. And I'm not even talking about the fact that it takes massive amounts of power, power and cooling to run quantum computers today.

So I'll wait on this bubble and see how long it takes for it to inflate. Our friends over at Al Kira have introduced two new AI tools. The first is the model context protocol server.

Everybody knows about MCP now and then network infrastructure assistant. These are designed to help make enterprise network management simpler and smarter. MCP links RA's network intelligence to AI assistance.

While NIA provides guided workflows for secure auditable actions with natural language interactions, IT teams can quickly ask questions, check policies, resolve issues, optimize performance, and they don't need any extra hardware or complexity that also leads to improved efficiency, security, and reliability across distributed networks. Now, Al, I know that network management is not something that you deal with on a regular basis, but this has AI and MCP written all over it. And since you're part of AI Infrastructure Field Day, I wondered if you might tell me why is it important that Al Qera has support for MCP?

Now, I think the, the MCP element to this is, is really acknowledging that there's an open standard for connecting ais together. That the insights you gain out of your network don't necessarily exist entirely by themselves. And so you may want some handoff to something that is, uh, doing ai, looking at other parts, AI ops maybe for your application or other elements of your infrastructure.

Uh, and I do see a lot of value in looking at the very complex environments that we are building and even enterprise data centers, cloud data centers, hosting locations. There's a whole lot of places where networks get incredibly complex because there's so many layers of abstraction there. And having an AI to assist you to identify the signal from the noise, find that the truth of what's actually going on.

We've, we've moved beyond the point. We're looking at log files and looking at config files is really a, a practical, manageable way of doing anything but the smallest of lab environments. So yeah, doing rapid movements from plain language, uh, description of symptoms to actual guidance for remediation, baking in some best practices.

I'm quite sure how AI like that is. Other than that, you've got a, a process that will run when you say you want it to, to run, uh, maybe a remediation to fix, uh, some routing protocol. Um, continuous monitoring and dashboarding.

These are the, the things that are highlighted as being the high impact use cases. Don't see anything newly AI in that. Having monitoring and dashboards is important.

Uh, compliance reporting documentation, that's auto created, real time configuration. Um, validation, policy validation. Again, none of these things to me scream ai, those, those scream software defined networking and centralized management, uh, getting user interfaces that are human readable, that's a little more AI like, uh, identifying the variability of the clusters of information that actually indicate a fault and being able to surface up an actual fault rather than all of the noise.

Those are things that require good analytics and maybe AI applications, particularly predictive AI applications rather than generative AI applications seem to be the, the right place to go with it. So yeah, I, I'm kind of a, a little skeptical on on how much value the generative AI part, the MCP and the AI AI agent part in here will play compared to what we've always had in previous, uh, environments that were software defined networking with programmability, how much the AI can take away the need to write that programming and write the automation, uh, remains to be seen. It's hope it's very successful.

While we're speaking about generative AI and AI generated code, Microsoft stopped a phishing attack that used a whole lot of AI generated code to hide its payload. There's a bunch of nasty JavaScript that was hidden inside what appeared to be a, uh, PDF document, but was in fact a graphics file and a, uh, an SVG file, uh, co-pilot security identified that this obscuring code was unnecessarily complex and unnatural, meaning this was created by ai, not by a human. And then defender was able to identify that the behavior of this, uh, SVG file disguised as a PDF file was not what it should be.

Uh, this is the cat and mouse of AI versus AI in malware. And researchers noted that AI makes attacks look more convincing, but it also creates telltale signs that users can detect. Uh, the case shows that both attackers and defenders are using ai, making it a crucial to prepare for more AI driven threats and AI driven detection.

Tom, are we AIing all the things? And is this gonna become izing all the things as we talked about before? I think it will.

And, uh, researchers were able to determine that this was a malicious payload because the file name had an M dash in it, which we all know is a sign of ai. Um, that, that's a shout out for my, my grammar friends out there. Uh, yeah, sorry about that.

Uh, this, this is a good example of one of the things that AI is actually going to be good at is this pattern recognition thing. So props to the attackers for wrapping JavaScript and SVG and PDF, that's a couple levels of detection deep. And the reason why they do that is because lazy scanning programs might pop the first file, see that it's a graphics file underneath and stop scanning right there, right?

Oh, it's not obfuscated. It's just something that I'm used to seeing. Uh, although I would wonder why you're sending an SVG file to somebody in A PDF and not some other file format.

But anyway, what's happening though is that the use of AI to create these very clever convincing attacks leaves all kinds of fingerprints all over the system, right? Like we know that AI is being used to create deep fakes, and we know that AI is being used to write more things and, and build images and videos and stuff like that, but the more we lean on it, the less convincing it is. I mean, think about all of the very first, uh, pictures that we've seen from AI image generation.

Like I still have a bookmark on my system of how to train things like chat GPT or, uh, you know, Dolly to create, uh, AI generated avatars. And by training it, you know, obviously it gets better, but it's still like you get the uncanny valley feeling, right? Like your brain knows that this isn't a person.

And so we are, we're finding the algorithms, but what's important is that the AI that we're using to scan those algorithms doesn't, doesn't ever get over the uncanny valley feeling like it knows the file type was way too complicated. The, the metadata in this does not match what a human would be like. We would expect to see a, a file that got sent out with like 14 different metadata revision updates, right?

Because that's how we write things. We write something, then we save it, then we edit it, then we save it, then we last for comments, then we save it. Like that's how files should look.

And unless we figure out a way to cause an AI to fake that ais will get something and be like, wait, just a minute. Something doesn't feel right here. And, and that's probably the best for right now because it allows us to set up these very simple, uh, pattern recognition systems at the, at the perimeter and say, okay, this doesn't look right and we need to quarantine this and analyze it.

And then you can, you know, take a little bit more time to understand what people are trying to do to you. I mean, this goes to the, you know, the old think all the way back to, you know, traditional antivirus programs that we used to run in like 2006 and oh seven. Um, the, the pattern matching from an AV program was designed to discard most of the junk right away.

It was the heuristics engine that was looking for virus like behavior for new strains that showed up that beat the patterns. And so you've gotta have that kind of defense in depth if you want this to work. So I think what you're gonna see is more people are gonna start adding AI pattern recognition, looking for these fingerprints of AI generated attack vectors so that what they can do is they can quickly discard that stuff before anybody has a chance to unleash the payload and cause problems.

Um, but you know, as long as you don't do something silly, like name your PDF, uh, Sarah Connor do pdf, I think you're probably okay. Microsoft is building fair water, which is a $7 billion AI data center in Wisconsin. And it's supposed to open next year.

It's gonna be 10 times faster than the top supercomputers and create up to 800 jobs, including a data center academy to help train local workers. It's designed for sustainability because it uses closed loop cool cooling. Yeah, it's designed for sustainability because it uses closed loop cooling and it offsets fossil energy with clean power.

Although those pesky environmental groups seem to be concerned about, oh, what is it? Oh yeah. High electricity demand of a brand new AI data center.

Al, do the people that are working on the environmental lobby believe that just using closed loop cooling is gonna be enough to offset all of the extra electricity that these power hungry servers are gonna be consuming? Absolutely. How could you question it?

Uh, this is pretty, uh, spectacular numbers for, uh, Mount Pleasant in, uh, Wisconsin. 3 billion, uh, initial data center, which is the one that'll be hopefully switched on early next year. Uh, Wisconsin's used to promises of big projects, uh, Fox gone, uh, that don't deliver what they expected.

So there is a little cynicism here. There's, uh, a reason for that. The locals are a little suspicious of these hundreds of thousands of Nvidia GPUs being pulled together with miles of fiber and as you say, closed loop, uh, water systems so they're not, uh, evaporating lots of water out and using up, uh, what would otherwise be fresh drinking water.

But of course, yeah, power is gonna be coming in large amounts of power. Um, Microsoft, Microsoft is saying that they're being efficient about it and that they're making long-term commitments to, uh, grid upgrades to avoid pushing costs onto the neighbors. 'cause, you know, high demand in one place means cost goes up for everybody around.

Uh, they're also working on a 250 megawatt solar project in, uh, Portage County. Uh, but the environmental companies are a little concerned about the capacity within the area to deliver this clean power, um, estimate so that this data center, this, this facility, this campus, I shouldn't really call it one data center because it is a pretty massive campus, uh, will consume, uh, the equivalent to energy of a million homes. Hmm.

There's an ongoing challenge for this build out of these AI data centers, and we've seen huge numbers of AI data centers being planned for build out around the world. Uh, it it'll be a challenge. It'll be a challenge to see that there is, uh, a value to business of value to humanity that matches up to the environmental impact for building all of these data centers.

This one fairwater in Wisconsin amongst, amongst others. Uh, hopefully that, that build out will progress, uh, successfully and particularly the build out of the energy resources required in order to deliver and to continue to operate this data center. Cisco has added AI agents to both WebEx and roomos to make collaboration easier.

Yes, that's right. We'll collaborate not just with other people, but also with an ai. Uh, the AI will schedule meetings for you, take your notes, summarize your discussions, uh, handle all of your routine activities and possibly even improve your meeting audio.

That assumes that there are remote people and it's all WebEx based now, uh, works with apps like Salesforce and Jira as well as with Microsoft devices. And there'll be, uh, tools added that make, hang on. Why is the AI becoming just more like WebEx, sorry, the polling tools and dashboards.

I like the deepfake detection. That might be a nice thing, but as we're gonna end up with AI agents managing our, uh, our meetings and other AI agents attending those meetings for us and summarizing the discussions that happen between the AI agents, where were they humans in all of this? Again, Tom, Uh, they're hiding.

Um, they are, they're currently trying to schedule a meeting, uh, and failing miserably. I like the idea of what AI can provide for these kinds of things because yes, the drudgery of scheduling and remembering to take the notes and the notes are too jumbled up. And I didn't learn courtroom shorthand, so I need you to summarize this and, you know, things like that.

It reminds me of one of the things that I've been thinking about recently that when we talk about AI agents, the conception that most people have of AI agents is, I want an executive assistant. When you think about the way that most executives are using ai, it's to replace a lot of their EAs functionality. Oh, I don't need to schedule that meeting.

I have somebody to do that for me. I don't need to take notes in this meeting. I have somebody to do that for me.

I don't need to worry about sending out in the meeting notes afterwards 'cause I have somebody to do that for me. And a lot of people who don't have a position that would require them, or basically necessitate having an EA will kind of use the AI functions for that. And I get it, right?

Like, you know, they want to be able to do all this stuff. By the way, adding polls to anything is a bad idea. Like I know that a Apple's latest iOS update allows you to add polls to a text message thread.

Nobody's gonna use that because I can't remember the last time I wanted anybody to give a poll about anything. Um, most of the time that I see a poll, it's a fake engagement thing, right? It's like, oh, you know, let's, uh, pick your favorite, uh, staple or, or something like that.

And, you know, racked with this emoji, like no, uh, deep fake detection. I can't tell you how many times that I've gotten on a, on a call, whether it's WebEx or Zoom or whatever, and thought, man, I wonder if the person on the other end of this line is actually a human being. Um, I know they're a human being though, because a lot of times they'll, they'll leave the filter turned on where they, they'll look like a cat in a court case.

Yeah, remember that one? Uh, that was actually, you know, that came up recently in a thing where we were talking about, you know, joining a, a, a legal proceeding and you forgot to turn off your cat overlay filter. That's how I know you're not AI is 'cause you, you make dumb mistakes.

I I think that these features have value. I don't think people are going to pay for them. So I think that that's the dividing line that we have to be aware of here whenever we're talking about all of these companies that are racing to add these AI features.

And, and I go back to my favorite, you know, bellwether on this. Would, for the people out there who bought an iPhone 16 last year specifically for Apple Intelligence, please step forward. I don't see anybody stepping forward.

Why? Why? Because as it turns out, people won't buy something for a feature like this.

They will buy something new in the hopes that that feature is included. Do they think that there's something valuable in it? Absolutely.

Being able to point my camera at anything and have my system tell me that, you know, that's what the kind of bug this is or that's what kind of leaf that is. Cool. I'm not paying $5 a month for that.

And if the hope from companies like Cisco and Microsoft and Salesforce and whomever is that by adding these virtual executive assistants to all of these platforms to do summarization and note taking and all this other stuff in the hopes that you're gonna get them to jump to a new licensing tier, I don't think you're gonna get the return on investment that you're hoping for. However, one thing that you can get a return on investment with is watching Tech Field Day. Because as you know, we are eat up with all kinds of great stuff and we have great things headed your way in the next couple of weeks actually.

So the first one of them is actually happening a week from, uh, tomorrow, uh, on October the ninth. I am gonna be having a special session with Microsoft Security. We talked about some of their copilot security stuff in this episode, but, uh, we're gonna be talking all about Microsoft Sentinel and some of the cool new things that they added to it this week as part of their Microsoft Secure event.

We're gonna have some live conversations with executives about the way that Sentinel has evolved over the years. We're gonna be getting some great demos about things like MCP and XDR and if none of those acronyms make any sense to you, you're gonna want to tune in on the ninth and check us out. And then the next week after that, um, our good friend Steven Foskett is off to the sunny world of Las Vegas for Tech Field Extra and NetApp Insight.

He is going to be having all kinds of fun hearing about all of the things that the NetApp ecosystem is doing to help accelerate the way that you are adopting new technologies. So make sure that you set your calendar for October 15th for that. And then the next week, Al, you're back on this side of the world again.

I am, I'm up for Cloud Field Day. We'll be, uh, up near San Francisco Airport and, uh, venturing across to Ville to see oxide computing. Uh, it's been a little while since we went to see them.

Uh, we'll hear from a couple of other great people. Uh, HPE and, uh, Fortine currently signed up a couple more that you'll see as well. That's October the 22nd and 23rd, and then Stephen Foskett takes back on.

He will also be back out west for AI Field Day seven the following week, October 29th and 30th. Uh, you'll find all of these, uh, sites, all of these, uh, events covered on our Tech Field Day com website and of course, tech Field Day, YouTube channel, uh, as well as catching up with all of our events. You can always catch up with all of the latest news here on the Tech Field Day rundown.

So thank you for joining us today, and do watch out for new episodes every Wednesday on YouTube, as well as in your favorite podcast application. The rundowns also streamed on Text Strong tv, and you can catch up with us either on Text Strong, as well as on various of the future and properties over time. We'll be back next Wednesday to talk about the IT news and the week that was, and until then for myself and for Tom Hollingsworth, as well as from all of us here at Tech Field Day.

Let's have a great week, and I hope you have an awesome week too.