Latest / Brian Thomas / Tech Friday w/ Dave Hatter - Sponsored by Intrust IT - 2/6/2026

Transcript

Tech Friday w/ Dave Hatter - Sponsored by Intrust IT - 2/6/2026 · Brian Thomas

• 0:00
  Speaker 1: The talk station.
• 0:03
  Speaker 2: It is six point thirty. It's Friday. Stay in the obvious.
• 0:06
  Speaker 2: It is time for tech Friday with Dave Hatter. In
• 0:08
  Speaker 2: trust It is Dave's company. He has been voted the
• 0:11
  Speaker 2: best IT company for businesses by the Business Courrier. I
• 0:14
  Speaker 2: trust the business career and I certainly trust Dave Hatter
• 0:17
  Speaker 2: and his team. So interest it dot Com to tap
• 0:19
  Speaker 2: into their awesome services. Welcome back, Dave Hatter. Happy Friday
• 0:23
  Speaker 2: to you.
• 0:24
  Speaker 3: Always good to be here, Brian, Thanks for having me,
• 0:26
  Speaker 3: and Happy Friday to you and Joe and all your listeners.
• 0:28
  Speaker 2: All right, everybody's seen the idea that artificial intelligence can
• 0:31
  Speaker 2: make clothed people look naked. But this is different. Apparently
• 0:35
  Speaker 2: chatbots stripper. Now you're gonna have to explain this to me,
• 0:40
  Speaker 2: you know me, I don't use made it and stuff
• 0:42
  Speaker 2: like that.
• 0:43
  Speaker 1: Yeah, you may be better off, Brian, you may be
• 0:45
  Speaker 1: better off.
• 0:46
  Speaker 3: So this is a story and wired and it's interesting
• 0:49
  Speaker 3: because you may have noticed in the media lately there's
• 0:52
  Speaker 3: been a lot of talk about this, specifically around Groc.
• 0:56
  Speaker 3: As a reminder, Groc is x ai a ka elon
• 1:01
  Speaker 3: Musk's chat bots. So when you think chatbot, when you
• 1:05
  Speaker 3: hear that word. Essentially, what they're talking about is large
• 1:09
  Speaker 3: language model based generative AI tools aka rock Gemini from Google,
• 1:16
  Speaker 3: Chat GPT from Open Ai, Claude, perplexity, et cetera. Now
• 1:20
  Speaker 3: it does a lot more than what I'm going to describe.
• 1:22
  Speaker 3: But you know the idea that if you go to chat, GPT,
• 1:25
  Speaker 3: you sit down, you type in a prompt and it
• 1:27
  Speaker 3: generates some kind of content text, audio, video, whatever. You know,
• 1:31
  Speaker 3: it's gotten much better at that these tools. And this
• 1:35
  Speaker 3: really gets to the heart of something you and I
• 1:37
  Speaker 3: have been talking about ever since AI sort of hit
• 1:40
  Speaker 3: the public's consciousness. Deep fakes which are could be audio,
• 1:45
  Speaker 3: could be video, could be text, but it's something that
• 1:47
  Speaker 3: is made up by the AI.
• 1:50
  Speaker 1: It's fake.
• 1:51
  Speaker 3: And you know, we've talked about the potential implications on
• 1:53
  Speaker 3: the legal system, on politics. But the latest thing that's
• 1:57
  Speaker 3: caused quite a stir is people have figured out how
• 2:01
  Speaker 3: to use these chatbot tools again think chat, GPT, rock whatever,
• 2:05
  Speaker 3: to take someone's photo and either a put them in
• 2:09
  Speaker 3: a bikini or b make them nude. And again this
• 2:12
  Speaker 3: story from whyre Google and Open AI's chatbots can strip
• 2:15
  Speaker 3: women down to photos, women in photos down to Bikinis
• 2:20
  Speaker 3: points out that this is fairly easy to do and unfortunately,
• 2:24
  Speaker 3: unfortunately are to stop. One of the terms you hear
• 2:29
  Speaker 3: when people talk about this is the idea of jail breaking,
• 2:32
  Speaker 3: that I'm purposely going to attempt to get one of
• 2:34
  Speaker 3: these chatbots to do something it's not supposed to do, right, yeah, like, yeah,
• 2:42
  Speaker 3: this is the latest thing. And you know, in many cases,
• 2:46
  Speaker 3: if someone just sits down and types in a prompt
• 2:48
  Speaker 3: give me this right, if that violates some rule, then
• 2:53
  Speaker 3: it'll say I can't do that. But what what people
• 2:56
  Speaker 3: have gotten good at is figuring out ways to spend
• 3:00
  Speaker 3: time working their way into what they want, you know,
• 3:03
  Speaker 3: step by step, trying to get to an end result
• 3:06
  Speaker 3: that would be denied if you just asked it straight out.
• 3:10
  Speaker 3: And I'm not saying that's necessarily what's happening here in
• 3:13
  Speaker 3: every case. I'm just saying it's really difficult to stop
• 3:16
  Speaker 3: people from jail breaking these things.
• 3:19
  Speaker 2: I just am led to the conclusion I just wonder,
• 3:23
  Speaker 2: don't these people have anything better to do with their
• 3:25
  Speaker 2: time day? I know that's a ridiculous thing to say,
• 3:27
  Speaker 2: but apparently they don't.
• 3:29
  Speaker 3: Apparently they don't, now, you know, and I understand why.
• 3:33
  Speaker 3: You know, if you're a woman and you have a
• 3:35
  Speaker 3: photo online and someone takes it and then put you
• 3:39
  Speaker 3: in a bikini or some other subjective thing or unfortunately makes.
• 3:43
  Speaker 1: You new it.
• 3:43
  Speaker 3: I mean, it's sad that this can happen, and I'm
• 3:47
  Speaker 3: not really sure ultimately it can be fixed. Again, can
• 3:52
  Speaker 3: they build these systems over time so that they can't
• 3:54
  Speaker 3: ever be jailbroken? I don't know, Brian, I really don't know.
• 3:59
  Speaker 3: I think it's a very difficult challenge. And then to your.
• 4:03
  Speaker 2: Point, a programmer, you're saying out loud, this is a
• 4:07
  Speaker 2: difficult challenge to get to that point. It's almost impossible,
• 4:11
  Speaker 2: is what I'm hearing.
• 4:12
  Speaker 3: Well, you know, maybe if the AI gets smart enough
• 4:15
  Speaker 3: that on its own, it could understand that it is
• 4:18
  Speaker 3: being led down a path to jail breaking to do
• 4:20
  Speaker 3: something it's not supposed to do. Maybe, But you know,
• 4:24
  Speaker 3: to your point, as a guy who spent a lot
• 4:26
  Speaker 3: of time writing code, could I anticipate every single way
• 4:29
  Speaker 3: that a user would do something that they shouldn't do
• 4:32
  Speaker 3: and that it would cause an error? It's really hard. Yeah,
• 4:35
  Speaker 3: it's really hard to do that, and that's why software
• 4:37
  Speaker 3: has bugs in it, right, you know, I know, it
• 4:39
  Speaker 3: never occurred to me they would do this one thing here,
• 4:42
  Speaker 3: and so it's a difficult challenge. I understand why women
• 4:47
  Speaker 3: are unhappy about this, especially like, you know, if this
• 4:51
  Speaker 3: was your daughter and some creep goes out and finds
• 4:55
  Speaker 3: a picture of her and next thing you know, there
• 4:57
  Speaker 3: are deep fake nudes of her floating or or again,
• 5:00
  Speaker 3: and suggestive photos and bikinis or whatever. I totally understand.
• 5:04
  Speaker 3: I think it's sad that this is a thing.
• 5:06
  Speaker 2: I think it could be embarrassing for guys too, right,
• 5:10
  Speaker 2: you know, I mean you could resize the junk and
• 5:13
  Speaker 2: embarrassed guys if you wanted, right.
• 5:15
  Speaker 3: Yeah, I mean you theoretically using these tools can alter
• 5:19
  Speaker 3: pictures in all kinds of ways, again, including nudes or
• 5:23
  Speaker 3: you know, could you try to frame someone with this?
• 5:25
  Speaker 1: Well, yeah, yeah you could.
• 5:27
  Speaker 3: It's this has been one of the latest things to
• 5:31
  Speaker 3: catch a lot of bad press. But when you when
• 5:34
  Speaker 3: you extrapolate out using these tools to take a photo
• 5:38
  Speaker 3: of someone or frankly, create a whole cloth photo of someone, yeah,
• 5:42
  Speaker 3: and then you know, put them into some kind of
• 5:44
  Speaker 3: compromising position, it's a real concern. And again, while groc
• 5:48
  Speaker 3: has gotten a lot of the the IR on this.
• 5:52
  Speaker 3: This is a story from Wired again pointing out the
• 5:54
  Speaker 3: Google and open.
• 5:55
  Speaker 1: Ai seem to have the same problem.
• 5:57
  Speaker 3: And you know it's it's just one of the many
• 6:00
  Speaker 3: challenges we face as a result of this technology. So
• 6:04
  Speaker 3: people should be aware of this. About the only thing
• 6:06
  Speaker 3: you can do to protect yourself from it is try
• 6:08
  Speaker 3: to limit the amount of photos that are out there
• 6:12
  Speaker 3: with you in them.
• 6:13
  Speaker 1: Yeah, good luck, which is difficult. Yeah, good luck.
• 6:15
  Speaker 3: And then understand that someone can potentially just create something
• 6:19
  Speaker 3: completely from whole cloth in and without a photo of
• 6:21
  Speaker 3: you to begin with, as this tech progresses.
• 6:24
  Speaker 2: So what a sad empathetic existence that person has who
• 6:28
  Speaker 2: takes the time to do this. We'll bring Dave patterback
• 6:31
  Speaker 2: talking about ways artificial intelligence can inflict damage. Plus stick
• 6:35
  Speaker 2: around comp It's six COVID sixty two. It's about kercittalk
• 6:39
  Speaker 2: station interest it dot com where you find Dave Patter
• 6:42
  Speaker 2: and you find them every week here on Friday morning,
• 6:45
  Speaker 2: meeting at six moving over ways artificial intelligence can inflict damage,
• 6:50
  Speaker 2: Unprecedented damage, says the headline of the zd net article
• 6:53
  Speaker 2: you'll have posted on LinkedIn on your account. Go ahead, Dave,
• 6:56
  Speaker 2: what's this all?
• 6:57
  Speaker 3: About Yeah, Brian, we won't have time to get into
• 6:59
  Speaker 3: all of these. I'm glad you mentioned that, so I'll
• 7:01
  Speaker 3: share this article. And I think that the headline is
• 7:04
  Speaker 3: a little hyperbolic, and I mean we just talked about
• 7:06
  Speaker 3: one of the ways they don't cover. This is more
• 7:08
  Speaker 3: of an IT focus for you as as a person
• 7:11
  Speaker 3: trying to defend your systems against this stuff. And I
• 7:14
  Speaker 3: would say, you know, every business needs to be thinking
• 7:16
  Speaker 3: about this, and I also want to start out with
• 7:19
  Speaker 3: these tools can be extremely powerful and productive if you
• 7:23
  Speaker 3: understand what they can and can't do, and you think
• 7:26
  Speaker 3: about some of the security issues that they point out.
• 7:28
  Speaker 3: In the cd in net article, a couple headlines are
• 7:32
  Speaker 3: a couple of topics. They say AI enabled malware will
• 7:35
  Speaker 3: unleash havoc. I'm just going to hit a couple of
• 7:37
  Speaker 3: these and come back to them. Agentic AI is involving
• 7:39
  Speaker 3: into every thread actor's fantasy. Prompting prompt injection AI tools
• 7:45
  Speaker 3: will be the new attack surface. Thread actors will use
• 7:47
  Speaker 3: AI to go after the weakest link humans. That's my
• 7:49
  Speaker 3: biggest concern. By the way, come back again, I'm going
• 7:52
  Speaker 3: to get into each of these a little bit. AI
• 7:54
  Speaker 3: will expose APIs too easily exploited. Extortion tactics will evolve
• 8:00
  Speaker 3: from ransomware encryption, and they mentioned in the next one
• 8:03
  Speaker 3: how contagion spreads to industrial control and operations. I don't
• 8:06
  Speaker 3: remember if we talked about this last year or not.
• 8:08
  Speaker 3: Range Rover Jaguar. Range Rover had a devastating ransomware attack.
• 8:12
  Speaker 3: They had some systems down for months, and the last
• 8:15
  Speaker 3: number I saw they lost something like two billion dollars.
• 8:18
  Speaker 3: People got laid off there, People got laid off with
• 8:20
  Speaker 3: their suppliers because their suppliers had nothing to do since
• 8:23
  Speaker 3: they weren't making cars. I mean, it was a devastating attack,
• 8:26
  Speaker 3: not necessarily AI related, but they're just drawing the nexus
• 8:29
  Speaker 3: of these things and the two big threats I think
• 8:34
  Speaker 3: every business really needs to consider around AI as it
• 8:37
  Speaker 3: stands today.
• 8:37
  Speaker 1: Again, there's a couple more in this article. If you
• 8:41
  Speaker 1: don't have.
• 8:42
  Speaker 3: People that understand what these things can and can't do,
• 8:45
  Speaker 3: the idea of hallucinations and so forth, and you can't
• 8:47
  Speaker 3: just assume anything that it creates as one hundred percent accurate,
• 8:51
  Speaker 3: that's a problem. And when you couple that with the
• 8:56
  Speaker 3: idea that I'm going to have an AI agent that
• 8:59
  Speaker 3: can go do something on its own. It's not a
• 9:01
  Speaker 3: human being sitting and typing prompts in and getting.
• 9:04
  Speaker 1: Output and doing something.
• 9:05
  Speaker 3: I'm going to let something run on its own and
• 9:09
  Speaker 3: do stuff. I'm going to give it credentials, use your
• 9:11
  Speaker 3: name and passwords so it can act on its own.
• 9:15
  Speaker 3: There is an enormous amount of risk in that. They
• 9:17
  Speaker 3: talk about that in here, and there are many other
• 9:19
  Speaker 3: articles about that. I'm not saying you shouldn't do it.
• 9:22
  Speaker 3: I'm not saying it can't work. What I'm saying is
• 9:26
  Speaker 3: if you think that you're just going to give this
• 9:29
  Speaker 3: thing a user name and password and let it start
• 9:32
  Speaker 3: doing stuff that that couldn't go catastrophically wrong, you need
• 9:37
  Speaker 3: to be very very careful and really understand all the
• 9:41
  Speaker 3: ins and outs of that. And frankly, if you give
• 9:43
  Speaker 3: it credentials and let it do things on its own,
• 9:46
  Speaker 3: have you secured it correctly so it can't be hacked?
• 9:48
  Speaker 3: And then the bad guys are now operating with permissions
• 9:51
  Speaker 3: in your system, the permissions that you gave it. So
• 9:55
  Speaker 3: that's something everyone needs to look at very carefully and
• 9:58
  Speaker 3: really understand what the risks are and make sure they're
• 10:01
  Speaker 3: taking adequate protection against that risk. But the biggest thing
• 10:06
  Speaker 3: I'm still mostly concerned about is as these tools progress,
• 10:09
  Speaker 3: as we just talked about in the last segment about
• 10:11
  Speaker 3: the bikini thing, it's easier and easier and easier for
• 10:15
  Speaker 3: bad guys to create things that are incredibly realistic, including
• 10:18
  Speaker 3: cloning someone's voice. For example, so you come into work,
• 10:22
  Speaker 3: there's a quote voicemail unquote from your boss telling you
• 10:25
  Speaker 3: to do a transfer of some sort.
• 10:27
  Speaker 1: Sounds exactly like them.
• 10:29
  Speaker 3: They got your voices or your boss's voice from his
• 10:32
  Speaker 3: voicemail or from some public speech he gave, or whatever.
• 10:35
  Speaker 3: They use a tool to clone his voice, a deep
• 10:38
  Speaker 3: fake voice clone. It's simple, it's free, and these things
• 10:42
  Speaker 3: are unbelievably good at this now, so you know the
• 10:46
  Speaker 3: idea that as a criminal, I can create text, audio,
• 10:51
  Speaker 3: and video that is unbelievably realistic. It will look exactly
• 10:54
  Speaker 3: like the real person. Could you get a call where
• 10:57
  Speaker 3: the hacker is literally using a voice cloning tool and
• 11:01
  Speaker 3: talking to you real time by typing a response in
• 11:03
  Speaker 3: and hitting a button and generating a voice. Yes, that
• 11:06
  Speaker 3: is a thing at this point. I think you and
• 11:08
  Speaker 3: I talked about the Ferrari situation last year where the
• 11:11
  Speaker 3: CFO of Ferrari almost was fooled by a voice clone
• 11:14
  Speaker 3: of the CEO of Ferrari. I mean they know each
• 11:17
  Speaker 3: other personally, they work together all the time. So I
• 11:20
  Speaker 3: think the average person does not realize how advanced all
• 11:23
  Speaker 3: this stuff is at this point in terms of this
• 11:25
  Speaker 3: deep faate capability, and every business should be training their employees,
• 11:29
  Speaker 3: you know, on the risks how to use this stuff,
• 11:32
  Speaker 3: the risks of it, and particularly cybersecurity awareness training that
• 11:37
  Speaker 3: includes something around AI, because I mean, we see it
• 11:41
  Speaker 3: growing and we've talked about the Grandparent scam and all
• 11:44
  Speaker 3: this stuff. Businesses are going to be increasingly targeted with
• 11:47
  Speaker 3: sophisticated attacks, not only from a technical perspective with AI,
• 11:52
  Speaker 3: but focused on the humans. It's one of the points
• 11:54
  Speaker 3: in this article, you know, fooling human beings into doing
• 11:57
  Speaker 3: something they shouldn't do using the advanced capability. So you know,
• 12:02
  Speaker 3: we can help. If anyone wants some help, call me up.
• 12:04
  Speaker 3: I'm happy to talk about this. But this is only
• 12:08
  Speaker 3: going to get worse before it gets better, because most
• 12:10
  Speaker 3: people do not understand how capable this stuff is in
• 12:15
  Speaker 3: terms of creating things that seem unbelievably realistic. Sadly, it's
• 12:19
  Speaker 3: believe it's a thing.
• 12:19
  Speaker 1: Folks.
• 12:20
  Speaker 2: Yeah, I can't believe you yearly can't linked LinkedIn dot com.
• 12:25
  Speaker 2: Let's look for Dave Hadder. You'll find him and you'll
• 12:27
  Speaker 2: find all the articles in information. One more and there's
• 12:29
  Speaker 2: a feature on your Google Gmail.
• 12:31
  Speaker 1: Five KRC the talk station.
• 12:38
  Speaker 2: It is six fifty one six two fifty five krcity
• 12:40
  Speaker 2: talk station. All right, what's the story on this Gmail
• 12:44
  Speaker 2: feature that you we need to turn off?
• 12:46
  Speaker 3: Dave hatter Well, Brian, you should know me well left
• 12:51
  Speaker 3: to know by now. My initial response is this headline
• 12:54
  Speaker 3: rather than if you use Gmail, you're going to want
• 12:56
  Speaker 3: to turn off this one automatic setting asapp should day
• 13:00
  Speaker 3: instead stop using Gmail. But good pud since it doesn't
• 13:05
  Speaker 3: say that, let me explain. Now this article is a
• 13:09
  Speaker 3: few months old, and again I'll post the link. If
• 13:12
  Speaker 3: you use Gmail and you plan to keep using Gmail,
• 13:14
  Speaker 3: I strongly recommend that you read this. And this is
• 13:17
  Speaker 3: one of the problems I have with all of these
• 13:20
  Speaker 3: big tech tech platforms. It's not just Google, right, So
• 13:24
  Speaker 3: many of these things are basically surveillance capitalism driven. You
• 13:29
  Speaker 3: are the product, not the customer. It's your data that
• 13:33
  Speaker 3: makes them quote free unquote for you, and when you
• 13:36
  Speaker 3: look at the ever increasing leaks of your data, the
• 13:41
  Speaker 3: ways this data is used in ways you would not
• 13:43
  Speaker 3: necessarily understand, and things like surveillance pricing. I just wrote
• 13:48
  Speaker 3: an op ed on this, which i'll have out there.
• 13:50
  Speaker 3: Maybe we talk about that next week. The idea that
• 13:52
  Speaker 3: because there's so much data about you out there and
• 13:56
  Speaker 3: these companies are capturing it, they're selling it to other
• 13:58
  Speaker 3: companies and so forth, I'm able to look at your
• 14:01
  Speaker 3: profile and say, you know what, I think Brian Thomas
• 14:03
  Speaker 3: would pay more for this than Dave Hatter. Looks like
• 14:05
  Speaker 3: he's probably got more money, or maybe he's more desperate
• 14:08
  Speaker 3: to get this thing, So even though we're on the
• 14:10
  Speaker 3: same website, you might see a higher price than me.
• 14:14
  Speaker 3: It's a real thing that's happening all the time. And
• 14:17
  Speaker 3: that's I bring that up because I think it's one
• 14:19
  Speaker 3: of the most concrete ways to help regular folks who
• 14:23
  Speaker 3: aren't tenfoil hat wearing nerds like me understand why you
• 14:26
  Speaker 3: should care about all your data. Because I still hear
• 14:29
  Speaker 3: why I don't care about that. You know, I don't
• 14:30
  Speaker 3: care about privacy. Second, nothing to hide, and I'm going
• 14:33
  Speaker 3: to come back to this article but I'm just trying
• 14:34
  Speaker 3: to set the stage for why you should care about
• 14:36
  Speaker 3: this and why I find this approach to things offensive.
• 14:39
  Speaker 3: So these companies will opt you into things. They roll
• 14:42
  Speaker 3: out some new feature that's beneficial for them, They turn
• 14:45
  Speaker 3: it on, and you have to figure out how to
• 14:47
  Speaker 3: turn it off, and often you don't even know this
• 14:49
  Speaker 3: has been done. Secondarily, even if you do, it's you know,
• 14:53
  Speaker 3: buried thirty six menus in and they tell you seven
• 14:55
  Speaker 3: times why you shouldn't turn it off. Dark patterns, trying
• 14:59
  Speaker 3: to convince you to turn off something that is beneficial
• 15:02
  Speaker 3: to them. And so to the heart of this article,
• 15:05
  Speaker 3: Google Gemini, that's their AI chatout platform, Generative AILM if
• 15:10
  Speaker 3: you will. They have now incorporated it into Gmail so
• 15:15
  Speaker 3: it can train off all of your emails. So if
• 15:18
  Speaker 3: you've been using Gmail for a long time, you might
• 15:20
  Speaker 3: have tens of thousands, hundreds of thousands of emails in there.
• 15:23
  Speaker 3: This thing is in there reading your emails, learning about you,
• 15:26
  Speaker 3: training purportedly to help you, right, because it's going to
• 15:29
  Speaker 3: help you schedule tasks and do things and compose emails
• 15:35
  Speaker 3: on your behalf and so forth. Right, But I'm going
• 15:38
  Speaker 3: to quote directly from the article for around of time.
• 15:39
  Speaker 3: Important message for everyone using Gmail. You've been automatically opted
• 15:42
  Speaker 3: in to allow Gmail to access all your private messages
• 15:45
  Speaker 3: and attachments to train AI models. Engineer Dave Jones shared
• 15:49
  Speaker 3: on X earlier this week. Again, this was a couple
• 15:51
  Speaker 3: of years ago, a couple months ago. You have to
• 15:53
  Speaker 3: manually turn off the quote smart features in the setting
• 15:56
  Speaker 3: menu in two locations. So first off, you have to
• 16:00
  Speaker 3: know that this has been turned on. Secondarily, you have
• 16:03
  Speaker 3: to know where to go. And then third you have
• 16:04
  Speaker 3: to go to two different places to turn it off.
• 16:07
  Speaker 3: And again it's it's the fact that they if you
• 16:12
  Speaker 3: have fully informed consent and you agree to use the
• 16:14
  Speaker 3: Gmail and you agree to use this feature, have at
• 16:17
  Speaker 3: it for an adult. Do what you want. I don't
• 16:20
  Speaker 3: use Gmail, and if I did, I would turn this off.
• 16:23
  Speaker 1: There you go.
• 16:23
  Speaker 2: That's the final point and we are at a time.
• 16:25
  Speaker 2: So heed his advice. Go to LinkedIn dot com find Dave.
• 16:28
  Speaker 2: How do you'll find these articles and you can read
• 16:30
  Speaker 2: the whys and the wearforce. Dave, thank you for what
• 16:32
  Speaker 2: you do for here on the morning show. My listeners.
• 16:34
  Speaker 2: Thanks to Interest I your company for sponsoring them. And
• 16:37
  Speaker 2: another word for Interest I interest dot com. Thank you brother,
• 16:42
  Speaker 2: have a great weekend. Don't go away. Corey Bowman up next.
• 16:46
  Speaker 2: Today's tough headlines coming up at the