Digital Nexus

13 | AI News talks Security and Risk

Chris Sinclair and Mark Monfort Season 1 Episode 13

In Episode 13, Mark and Chris discuss a wide range of topics, including the evolving role of AI in security, blockchain, and the potential of autonomous vehicles. They dive into insights from SXSW Sydney, Tesla’s AI and automation advancements, and the impact of Google’s nuclear power investment to support data centers. Key discussions include AI’s vulnerability to hacks, the benefits of AI in real-world applications like blockchain payments and superannuation systems, and the ethical concerns surrounding new technologies. The episode concludes with reflections on future AI developments, the rising influence of tools like Copilot, and the ongoing need for careful security and data protection in AI applications.

Timestamps:
00:00 – Intro: Kicking off Episode 13 and key themes on AI and security and recap from the week
05:03 – South by Southwest review: blockchain projects and AI integration in payments.
06:32 – Build Club at SXSW, their raise and recognition for their founders
11:40 – Google’s nuclear power investment to support future AI data centers.
15:19 – Tesla’s new automation technologies, including robot taxis and autonomous vehicles.
21:55 – SpaceX rocket recovery innovations and AI in space exploration.
23:23 – AI reasoning limitations: insights from Apple and Francois Chollet.
27:17 – Gingko Bioworks exploring AI and biology
30:00 – Pyramid - new text to video tool
32:37 – OpenAI’s Copilot and the importance of AI security in organisations and other vulnerabilities
40:37 - Opera browser with AI
44:17 – Closing thoughts on AI’s future, security, and upcoming projects.

Other Links
🎙️our podcast links here: https://digitalnexuspodcast.com/
👤Chris on LinkedIn - https://www.linkedin.com/in/pcsinclair/
👤Mark on LinkedIn - https://www.linkedin.com/in/markmonfort/
👤 Mark on Twitter - https://twitter.com/captdefi

SHOWNOTE LINKS
🔗 SIKE - https://sike.ai/
🌐Digital Village - https://digitalvillage.network/
🌐NotCentralised - https://www.notcentralised.com/

YouTube Channel: https://www.youtube.com/@DigitalNexusPodcast
X (twitter): @DigitalNexus

Welcome back to another episode, we're at episode 13. Pretty excited, we got a lot of things to talk about. 13, 13, 13. We had a whole lot of stuff around like the fake AI and what that kind of means and stuff and like the hacks that come. We spoke about co-pilot as well and like there's things that people have been able to show the black hats out there in terms of hackers have been able to show that you can get into co-pilot. So be careful. A lot of all the latest news on Tesla. Tesla. That's a massive update. Not iRobot, we robot. I'm coming news in terms of open air, and chat GPT, a lot of open source new tools that are out there in the market. Listen in episode 13, let's go. Be there and be square. Welcome to Silly Fridays folks. It's Friday the 13th episode as Chris drops stuff very spookily it's episode 13 of the Digital Nexus podcast and I'm mark founder of psych.ai co-founder. Sorry. I should say the code The got to be Mortified now you wouldn't be mortified. He does not care You know, I mean even before AI we had AI we had our Turo intelligence So like it's always been that which has been great. Um, but yeah co-founder of psych.ai Co-founder of the Ozdeaf eye Association, which is going through some interesting Revamp type stuff will be excited to share with you as soon as we get that I'm here with my co-founder of this podcast Chris Sinclair Hello Yeah, man, not no cost no co-founders and other than this podcast but definitely partner with Digital Village as we've talked about many times before Shout out to the digital guys everything to do with you want to find specialists Villages digital stuff. Yeah, it's them. Well the guys it's great. Um, I do a lot of UX Another work with many clients looking at AI platform development It's everything from digital ecosystems through the stuff that you get into the that you put into the market So yep, and just to not confuse things Chris is the blonde one behind me. So just you know, you might get confused I used to have a beard. That's the yeah, it's weird and then I bring back Yeah, for Movember, it's so well, that's a mode or is it be January. I don't know be January be a good January Anyway, the AI will figure out how to do it Anyway, the AI will figure out something better Usually we check on how each other's doing. There's a whole lot of like I think we're both in Project mode and stuff for us. Yeah, it's been it's been intense, but I've really happens when we're in need for this project running We're in a huge creative phase where we've done all this in Amazing research and we're gonna talk about how we did some of that research in a moment a small part of that research phase insights But yeah, we're in that back end where we were taking that research now We're bringing it to life on what the solution should be for the customer. So But it's the fun part of the whole equation. I love it's fun, but stuff always happens like when there's a big conference kind of week and as much as like we want to head down and check out the Music the media that the cool interactive stuff that goes on a South by Southwest. It's just been so busy I had my talk the other day which was on blockchain. I want to know all about this How did it go blockchain and how? Elements of blockchain can help the future of the digital world that we move to right? Yeah, and I think one part of that is like how it helps AI so I spoke about that I spoke about the blockchain projects we're doing around like there's payments kind of system that a lot of us are just unaware of especially those that rent and not realizing that a lot of it is gonna be in terms of how things get paid and you and I don't have to worry about it because we just pay as Pernormal, but the underlying rails will be on this much more efficient Transparent as well in ways that it needs to be and also being able to have Compidentiality where you need to have it and being able to select that Running on blockchain tech and so that's like one thing There's been other projects that are being seen out there that you would not be able to have so there's a whole Superannuation kind of change where you now have payday super so the ability for You to instead of waiting for super to go from your employer to hit your accounts It is there the same day and especially if you're an SMS F and and stuff you can you can do more with that So there's interesting stuff that is happening because of blockchain so I think it was really interesting to just like talk about it and just get people to realize that a Lot of the the hype the memes the bad culture the crypto bro culture There's a lot of really a bro real interesting and there's a eyebrows as well, right? There's a lot of really interesting useful stuff that happens Behind the scenes that people aren't aware of so it was great and we got a really great audience and stuff after so that Was South by Southwest for me. I'm gonna try to head back today. It's Friday But yeah, we'll see did you did you get to attend anything else and at South by's was mainly being about your talk And how did it go? How did it go to talk went? Well a lot a lot of good questions and stuff that well, there's only one question from the audience We only had a half an hour. We stretch it to 40 minutes and stuff, but there were some good conversations that happened afterwards and Yeah, I was with Rob Alan from Australian payments plus, you know, well Kate Cooper for a Zodiac custody and then also Amy Rose goody from She's a CEO of the Digital Economy Council of Australia, which was formerly blockchain Australia It's a really interesting crowd that was there. So that went really really well Good turnout. I haven't checked out anything else. I know there's a there's like the AI hub stuff that Accenture is doing build Club is doing a bit of stuff So I'll jump to that one just just I spoke to a few people about it They've been very underwhelmed again this year from South by but what has been Tremendously awarding for those have attended is mainly there've been a better focus around the networking aspect of things. Yeah connecting people up about a meet Interesting other people whether it's in the same field or other fields. So that's been that's that's an improvement on the overall experience Which is great. Absolutely and speaking of South by Southwest build Club These are some highlights from they did on they've got a hackathon that's going on right now I listened in last night to well, you know, you're watching this in the future So it's all gonna be decided by then but given that it's Friday and we're in the past on Wednesday Let's just jump around Marty gotta come with me Marty So basically they they did they set what the challenges were the ideas at this thing in South by Southwest last night being Thursday They had a group of builders come in To an online virtual gathering form teams and now they're gonna go off and help You know build something and show something within like over 24 hours. So I think like by Saturday or something they have to produce video and some examples of what they've created for the hackathon using AI and there's been some folks like Build ship as well as relevance AI and a few others that are being here's some credits Here's some stuff that you can use our software to do so really interesting stuff And was there any anything that people showed off in the event itself? We have no to the weekend No, we got away to the weekend. So they've only just started there was only last night that they've formed teams But they could effectively build from last night. So I'm sure that we'll highlight and you'll see on LinkedIn when you're watching this one You'll have been able to see who the winners were and what it was that they built But yeah, really interesting stuff will put those in the show notes. Hopefully if the if the yeah by that. Yeah, yeah Actually, yeah. Yeah, we might be able to as well as Interesting to highlight apart from South by Southwest. It's all other things on build club So they've done really well and he's gotten through to the Forbes 30 under 30. So amazing stuff applause again Well done not that was real applause or maybe it'll be AI generated by the time this is edit Yeah, so Annie so young so amazing doing really cool stuff with build club from a vision That was just a side hustle when she was working at a VC firm to now being able to go into it full-time because At the same time they've also been able to raise a bit of cash 1.8 million So for her, I think it's Tom and Vincent that basically started this I've had Vincent on the Oz DeFi podcast part one Have been able to get to part two because it's been so busy but really great insights and stuff that he's got there Around design thinking he's been a builder in the space. He's been in the AI space for a long time even before Generated AI bit was a thing a fellow design thinker a valid a fellow design thinker Um, so another thing to highlight was that they raised the 1.8 million and yeah, they basically put it out on notebook LM So notebook LM is here talking about it It's a pretty ambitious goal. And I think it's amazing. There were, I hadn't looked into it, but there were the government awarded, I think three or four sites around Australia to be AI centers. Now I haven't seen anything come out of that. I even got in touch with a few of them. Talking about centers as in actually. Like centers for learning and educating. Harves in education, yeah. Yeah, yeah. For, you know, existing places that wanted to create like a program and a few of that, they got like, I think they shared 1.7 million or something like that. So a good lot of money. How many people? I don't know. I have no idea because there's nothing in terms of like, what it is that they're doing. I haven't seen anything come out. So hopefully the money's being put to good use. Yeah. But here clearly, even before getting funding and stuff, there were already, you know, getting around to different cities and bootstrapping their way through to getting funding now to actually do something. So it would be really interesting to see what they do there. But yeah, great that they highlight, you know, they're using notebook LM to create this podcast. Speaking of notebook LM, the bug there with that was that you only had those two presenters. You can't really control it, but they're coming out with something where you can control it. Really? Yeah. It's an update to the Google and it's gonna allow the more customizability to the actual people who are talking. Correct. I wonder how they're gonna manage that. I don't know, but they're gonna find a way. It's Google. Life finds a Google. Life finds a Google. So I thought that was like really interesting to highlight and then there's a bit of other news that we can flick through. But what do you wanna do? Do you wanna go through some of the new stuff or do you wanna? Mate, you're on a roll, keep going. I'll keep going. Another thing to highlight is this thing from Google here. So they're going to basically go nuclear and this is not figurative. It is literal, literally. So this is, I found a few pieces on this. It's been a big topic throughout the media over last week. So Google has backed Kiroz Power over international. Kiroz, is it? Kiroz, I believe it's Kiroz. Apologies. They're backed Kiroz Power to build essentially that over the next 20 years, over seven reactors. With their means obviously to also feed the local grid but essentially allowing them to tap into as much power as they need as well. Yeah. Some stats came out around some of the data centers that Google is currently managing where essentially they've been using almost up to a hundred megawatts of power, which is not an unsubstantial amount of power but they predict that over the next sort of 10 years that could increase to more than 1000 megawatts or over a gigawatt of power. So there's data centers consuming an extremely large amount of energy across the fact they're doing the cloud or they're based their normal data, AI and the works. So there's a real need to be able to, I guess, expand on how they pull energy to service that infrastructure in a cleaner sense, if you could argue that word. Absolutely. I mean, and then they highlight that by 2030, using Psyche to get the summary because you know, lazy within use and stuff, they reckon power demand is gonna grow by 160%. And the sector could accumulate 2.5 billion metric tons of carbon emissions. So that's a heck of a lot that we can do now that, you know, we have this environmentally kind of focus society and we have carbon credits. I mean, you know, there's a whole lot of economic potential benefits that are gonna be done there. Small modular reactors, they speak about that as well. So this is, they're gonna do some stuff in collaboration with Energy Northwest and have these SMRs, which are considered more effective and efficient than traditional nuclear reactors. So that'll be like a key technology for achieving that 100% clean energy. And energy is an interesting one because blockchain and AI have gone, AI is going through currently in terms of, well, the energy production, it's so bad and stuff, so what, we should stop using AI. Blockchain was going through before because it consumed so much power with everyone using GPUs and, you know, to power the mining, but more than 50% of blockchain is now using clean energy. So it just takes time, but it gets to a point where the energy factor isn't an excuse at all because in fact, in some places, in some data centers, using blockchain, the data center places have forced being, you know, they've innovated to become more efficient, which is actually a benefit for all data centers and stuff. So innovation begets innovation. It's just like when people go,"Well, why are we sending men to the moon and like people out into the, you know, into space?" And yet we get all these innovations that come from us going out and exploring. So it might not be, I'm sure we've got to be pragmatic and stuff like that, but we need to keep exploring. As long as there's more good that comes from things rather than negative, I think it's all good. So, yeah. I mean, the two biggest sectors of innovation, if you think about it from a technological standpoint is military and secondary being space exploration. Both of those are responsible for astronomical amounts of technical innovation, even leading all the way down into AI and machine learning. Like the rules of that come back to, you know, how do you identify and break codes? That's kind of where machine learning came from. Yeah. So it's innovation begets innovation exactly as you highlighted. Yeah, innovation, that is gonna be the, I don't know. Speaking of innovation though. What do you got? Like we can't not talk about what happened last week. What happened last week? Mr. Elon Musk. Oh, okay. You wanna talk about Tesla? Tesla's massive announcement. I mean, this has been all over the media as well. It's, I think if you haven't heard about it, you've probably been under a rock. But obviously Tesla had a huge announcements over the last couple of week in terms of some new technologies. And a lot of it to do with AI, a lot of automation, probably a better word to utilize in this space. But we've got the release of their new robots, their robot taxis, their automated vehicles. I'll even bring up a video here. Just like you can see here the big, nope, we've got ads popping up. Was, this is. Brought to you by Squarespace and NordVPN. And NordVPN into Squarespace. This is a video up here of their automated vehicles. They all look like model wise. I think they got some model three style variants. And essentially these are, these robo attack, all these robot vehicles, they're not these other taxis, these autonomous vehicles. And essentially anyone can purchase and own these vehicles. They come and they pick you up. Or does the vehicle own you? Or not there? Potentially. Or not there. But one of the most interesting things about this is that you can then forward on all the monetization of these vehicles as well. So you can literally start to, when the vehicle is not in use, it can then drive away, go pick up someone else and that person can then pay for the usage of that vehicle. So your car is no longer sitting there doing nothing dormant. Lazy. Those lazy people from the early 21st century, like having their cars just sit there. Oh, it's fantastic. Yeah, there's a really good ad. I can't remember what it was, but it was, or who came up with it, but one of those European clever ad agencies and stuff showing how much space that we take up when we're in cars and what they had. Like you see cars out here, for example, in traffic and people saw all the cars banked up yesterday because of the tragic, the bridge accident and stuff. So knock on wood there, but like they showed people. Yeah, in Sydney, we're in Sydney folks, showed people as cars like moving around, instead of the cars, they just had people standing there in position of where they would be in the car. And you see these people kind of like moving around through the city with so much space between them because that's literally what we've got with the cars and stuff. And I think it's a great idea if we can find more efficient usage of these things. We'll get pushback though, very likely you'd think from the vehicle manufacturers, it's to the guests and all that stuff. We're in the future, we're pushing for, they're not gonna be able to stop this stuff anymore. And we have to, but no doubt. This is a friction that always happens when there's an incumbent industry. So people are seeing it with emerging tech. So yeah, really interesting stuff. And the future of like urban impact and new products, I think they also discussed that in the podcast by converting a lot of parking lots to green spaces. This is it, this is the big thing, especially with these automated taxis now, you're able to pull back on the amount of infrastructure of roads that you and I would generally need when we're driving our vehicles. You're able to optimize flow of traffic through cities. It even improves how you get, pick up items and distribute it out to, whether it's transportation of goods and services, these vehicles can tailor to that as well. It's just like the clean energy opportunities here are like beyond imaginable, it's fantastic. I think we need to get to a future like Futurama, where it's a whole series of tubes across the city. We don't even need vehicles. You build a city on top of our city. Yeah, like you jump into a tube and it takes you straight to your office and stuff. Now you might have a little bit of a headache, but there'll be a pill for that. So you'll feel a little bit jelly, so that's fine. But yeah, interesting stuff that we're heading towards. So they predict, I think the autonomous vehicles by 2027, I think they were saying, and everything else around 2030. You see here, up here, here's the robots coming out. I-Robot. Yeah, essentially marching out in an I-Robot formation. I believe there were-- Is that the T-1000? I believe someone was saying that they were actually, these were controlled by-- Xbox controllers? Yeah, some of them were probably the Xbox controller, so they're not actually AI infused yet. Probably more for the performance to make sure they were optimized and-- That's just a dude in a suit, a really skinny dude. Exactly right. But the thing is, it was the way that they were still able to move, which even with a controller, we haven't had two legged robots comfortably moving about in this human manner at all. So this is incredibly exciting to see. They've released possibilities of the cost of these. We're looking at probably 50 AUD, 50 Australian North. I think it was like 30K Australia USD for the ownership of one of these. And I think they're saying around 2030, if I remember correctly, as well that these might come to market. But that's not far away. That's like five, six years we're talking about automated-- That's crazy. AI driven robots living in your home, helping you do your day-to-day chores, keeping your kitchen clean, delivering your groceries. Yeah. This is happening. This is the future. Yeah, I think there's gonna be a lot of people really pushing for this stuff, a lot of people rallying against it. I can see protests in the streets, I can see paranoia at something so different. As long as we show the benefits, then-- There's Elon Musk's creepy face. He's like, "Yeah, it's happening."(laughing) That's Elon Musk in dreamlike state. So, I think it's really interesting, the future of where we're heading to and the economic and safety benefits are gonna be really cool. So, great stuff from Tesla. They also had the rocket catch and stuff as well, instead of having to land and have feet that extend. They also had the rocket catch where the arms of the thing that usually hold up the rocket were shown there, but not AI. So, are we gonna go to it? I'll bring it up, because I think it was phenomenal as well. So, this happened earlier this week. I think it was on Tuesday or Wednesday. SpaceX catches essentially a-- An ad. Part of the, I mean, the idea is that when rockets take off-- Pay for YouTube, folks, pay for YouTube. Yeah, I shall wait for that. So, a KO, showing a KO, which is great. The irony. Irony Fridays. I'll get straight to the rocket. SpaceX. There you go. There we go, look at that. So, essentially here, like with rockets, when they take off, they have their two big rocket packs or less and they get dropped. They have huge amounts of resource to then go and try and pick them up and bring them back. If they are able to, deep within the ocean. Previously, Tesla, previously SpaceX has been landing these rockets mostly successfully with legs on the ground, but some of the issues that you have with that is the damage to the cement or the concrete. When it lands, it's literally cooking the ground, which is causing stability issues. It's then having to rebuild the infrastructure for the land that is underneath the landing pads. They were trying to explore other ways of capturing these rockets after they've been utilized to launch people off into space and boom, now they're just catching them in the air. That's a good way to do it. There's a lot of other benefits to this about you can now start landing these things in areas where there might not be very comfortable terrain. So, you could build this infrastructure somewhere else, whether it's out in space, have those rocket things grab them as people come and land in different atmospheres, different terrains, different planets, like it's the different possibilities here now, endless, it's fantastic. I think it's gonna be really cool seeing how all of that that works and stuff. So keep an eye out for that. Another interesting thing if we just like switch back is I've already got the key points there, that's fine. I was just gonna show something, but yep. So we've got Apple here. This is the AI grid, definitely a podcast to watch out for, but they highlight in here and I'll play without the sound and stuff, just so you can see some of the things are going through. So it's a very like much more reasoning kind of thing that's being shown here. It's much more paper based, but it's about reasoning. So the idea is that Apple, and we've seen others talk about this and I know that people might take this as a negative, but the limits of like AI reasoning and it not being able to do reasoning as well as we thought. So they talk about things here from this podcast. So when it comes to, for example, let me just get my notes here, just play some music in the background while we do this.(humming) Knowing the automated AI is gonna like pick up on this and make this a short, which is stupid, don't do that. But basically the points were that when it comes to like reasoning, there's limitations to what is actually happening with reasoning. So if it hasn't seen a problem before, it's not gonna be able to do anything kind of further with that. If it's not, if it is able to do something like say on a test, like here's a problem, solve this, it can usually do it really well when it's seen that problem before. So Francois Chollet, AI researcher, he's done some stuff in this space, Apple's now come out with some official phrasing and reports about how they see it as well. Now it's also sensitive to changes and there's these tests that they call GSM symbolic benchmarks where what they're doing is they change the names and values in test questions to see how big the performance drops. So if the problem was George, Tom and Jerry go to blah, blah, blah and have to row a boat across the whatever, changing the names of the characters, but the test is exactly the same, a lot of AI can't figure that out. Because it's not, it now sees it as a new problem. So this is a thing that is a problem when it comes to like AI reasoning, that you can't just let it, it's not gonna, it doesn't look like it's evolving basically. Like the way that we construct large language models right now, they haven't been able to figure out a way to get beyond that. Now the interesting thing is that that's a benefit I think for having humans in the loop. A lot of people, you know, for them is like, AI is not good enough. Until they solve it. Until they solve it at least. AI for some people is like, well, it's useless until it actually solves something, but it's like, by the time it's actually doing everything autonomously, it's useless until it's autonomous. If it's autonomous, you can say goodbye to some jobs potentially. So I wouldn't welcome full autonomy just yet, even though we'll probably will get there. You know, there's this period of time where we can really learn to work with it. And this is an opportunity to embrace the tech and understand it because through doing that, you can better position yourself for when that future does come that, you know, these things do have AGI and it's autonomous. But it is a benefit right now, especially if you're a consultant, it still means that whether it's consultant or you're working in house, that you have to know what it is that you're doing with the AI models. Like they're not going to be able to figure out and just reason on their own. There's still a lot of steering that you need to do. It is getting easier and there's more autonomy in terms of how people do the steering. But I think it's really interesting that Apple is also highlighting this kind of stuff as well. So that's definitely one to highlight. I've got a few others here. How did you want to do this? Yeah, there's a couple of new things that have come out to market. Let's touch on this. We're showing online. This is an interesting one. This is a biotech company which has called Ginkgo Bioworks, which has a really interesting exploration into mapping of our, I guess, the biology of our genomes. And they're coming up with essentially the chat GPT of this biotechnology wave. And so they've just done a few announcements over the last week, they're stating here, we're basically making this ecosystem by which Ginkgo becomes the tool provider to the entire industry, AKA the biotechnology. What we're doing here is basically making a model API akin to OpenAI or Enthropic, so that others can now build great tools on top of them to access these models and predictions. And these predictions able to understand, you know, like the biology of humans, animals, creatures, worlds, plants and stuff like that. And so really interesting tools coming to the market. We talk a lot about, you know, large, LLMs, large language models and, you know, video editors and, you know, data tools around the AI space, but the amount of work that is being done behind the scenes in the health industry and the biotechnology industry, even in infrastructure and management, there's like really interesting tools that are coming about and improving how we're able to do research, how we're able to identify even like issues in health far in advance with predictive models. It's really fascinating what's happening in that space due to this whole, you know, dynamic of AI coming into the, into the hemisphere. I think, yeah, it also highlighted the future of drug discovery there in that article. Yep. So saying things like they, well, I think that they interviewed the founder there, where he was talking about envisioning a future where every company developing therapeutics will have these AI tools, either by investing heavily in their development or by partnering with, with Ginkgo. So really interesting to see that they did call it the chat GPT moment for the space, as you mentioned. So I'm really interested in interested in seeing what others in terms of different industries, not just within biotech, but in other industries where they can be inspired by having, okay, what is their chat GPT moment? It's like when notebook LM came out for podcasts, it was like, wow, this is the, could this be the chat GPT moment for podcasters? We'll, we'll see. But there's definitely a need to have that fine tuning idea in mind when it comes to how do these tools work for your specific industry. So it'd be interesting to see where they, they end up with this kind of stuff. So we'll have to keep an eye on that one. Yeah, very nice. Couple of smaller things in that have launched over the last couple of weeks. We talk a lot about these video editing businesses and organizations, AI tools that have come to market. A lot of them are, you know, they're owned by these companies. This open source one, which has been pretty interesting, has just done a few updates called pyramid. And pyramid has released a couple of videos just around what it's capable of doing. This is like text to text to video with some realistic video content. You can see here. Nice. There's some really like quite realistic and very minor issues. You can see little errors popping up, you know, like with the along the legs and the feet, some of the transitions here, but just the landscape is beautiful. The lighting is beautiful. It's ability to, you know, to render movement in a meaningful way is really cool. So they're able to, you can see here, there's the car sort of like leaning into the road slightly there as it's driving by. But the fact is that this is an open source tool. So anyone can utilize it, download the API, start running things locally on their systems, you know, and then tinker and improve how these other tool works as well. So that's pretty cool announcement just because like, oh, there hasn't been a lot of open AI, oh, sorry, open AI, open source tools available for video. Well, speaking of like the fake stuff, there's been a bit of stuff on fake history that's being created there. If we can, how do we switch? All right. We're going to work out the transitions and stuff a lot better next time. Sure. Which one cut? Okay. We're going to cut better next time and stuff, but yeah, look, there's new tools here folks, new tools folks. This is what happens and we'll sort out the order and stuff like that for the future ones. But yeah, fake, fake, you know, stuff that's on there, it could lead to like some negative things because people could create like fake history photos, as you can see from this article, where you've had things like, hang on a second, I think it's this one here. So for example, we'll play this short where it's been doing the rounds where it's like, this is how the ancient Egyptians made the pyramids. And it's this, I mean, if you didn't know, you'd be like, whoa, look at how big these people are. It's like, hang on a second. This something's on right there. Oh, it's just playing on those people with these. Oh God. You can see people falling for this kind of stuff, right? Like it's going to happen. Unfortunately. Giants existed, didn't you know that? Giants did exist in my head. But another thing that was interesting was this piece here from the Black Hat podcast, which is all about the, it's a conference that they have every year. And I think they maybe do multiple things throughout the year, but hacking into things. And in this track, it was all about co-pilot. Co-pilot is not as safe as you think. So just need to be quite careful with how you use things in there, because as the Black Hats have shown, there are ways that you can actually get into it. And so one of the things that I need to highlight, so let's just take a look at Sykes, give me three key points from this podcast. And we've highlighted co-pilot down here. So I've just grabbed the whole transcript there. And for me, one of the key takeaways is that anything where there are actions that the AI can do on your behalf is vulnerable. What that means is that where the AI can, hey, you can send out emails for you. We can go look up files. I mean, just looking up things is certainly one thing. It can't turn off the sirens, which I don't think maybe the filter is, we'll take off the sirens. Yep. We'll do that. But the thing that it is interesting there is that it's going to be this world where you need to be careful about what you put into the co-pilot. I just, it just, it just proved, it wasn't really happy doing that. Can you fix me up? We're having a great day, man. There we go. Yeah. Just needed a little pill and it's okay now. But no, the interesting thing there is that, you know, for the microphone folks, for the microphone. So this is episode 13 and things are supposed to go wrong. It's spicy, spicy episode 13. But no, the thing that you need to remember is that anywhere there's actions with co-pilot or other tools, it's vulnerable because we're in a world now where you don't have to code to be able to get AI to do stuff or to get programs to do stuff. You just chat with it. So think about how you can socially engineer people to convince them that, yeah, you are an employee of this company and stuff we're doing. And they show that you can do very similar things to the AI by saying certain phrases, key words and stuff, what they call incantations. They're able to bring about some really interesting insights that the AI is able to share. And because it can share those things, they can do normal hacky type stuff where you get an email, but embedded in the header of an email is the instructions that only co-pilot can understand. So you just gotta be like really careful. If it's read only AI and it's reading and giving you stuff that it produces from just reading stuff, but it can't write, that seems like it's a little bit safer, but especially anywhere where there's like AI actions and that's exposed to like the outside world somehow, like you do have with co-pilot. I mean, they even had something where co-pilot, and this is just current time in terms of writing or at the time of the video, it could not tell if there's a Chris Sinclair from an entity outside of Digital Village that makes up its name to look like it's Chris Sinclair from inside of Digital Village, then it doesn't know. It doesn't know the difference between the two and they showed that with this. So just need to be mindful. He goes through in detail here. So I think it's a really important video for security researchers just to be aware of. That's a good call. So has anyone used co-pilot? Yeah, a lot of people use co-pilot for doing summaries of emails and things like that. I think it's a bit, even though they can do it, it's like early stage can, limited kind of features with their designer kind of tool and with PowerPoint. And they've done a lot of cool updates to the over the last couple of weeks, even I think we talked about some that came out last week for co-pilot, but yeah, I think it's good to be conscious. I mean, it's not only about co-pilot though. If we think about it, it's any AI tool that you bring into your organization, that you bring into your own life, education and being conscious of what it is you're doing with these tools and where you're doing them is super important. I know AI has obviously got a massive magnifying glass over these security risks, but even prior to that, the caution that was required even just to jump onto your email is always been super high. Just obviously that lens has now shifted to something that is driving a lot of data and a lot of decision-making for yourself, for businesses and for the world. So being conscious and being educated is really, really important. And now we rattle home around consultation and having people, specialists come in and supporting organizations when it comes to the use of AI, but because of these very reasons, if you ever watch this for a video and you highly recommend it, especially if you're interested in how to integrate and understand the security risks of AI within your organizations, the risks associated with not knowing this stuff are significantly high, especially if you have a lot of data that is high risk. A lot of personal information for your customers, a lot of, could be IP that you were talking about and not being aware of how these tools impact all that decision. It can have a huge risk across your organizations. And that's why consultation is really important because these people, these specialists know about these problems. They know around, they can learn about your needs and your wants and then put the right tools in place to lower that risk and therefore, and still provide that level of innovation that your business needs to move forward. Agreed. And speaking of AI risk and stuff, another thing to highlight here was that there's a new Gmail security alert. For 2.5 billion users, as AI hack was confirmed. So it's an AI driven Gmail attack. So you can check out this article here from Forbes. Basically, they highlight a few things like the number that was being called while they were speaking. It leads to, for example, you get a message call certain number and it seems like it's Google and it leads to Google business pages, but there's ways that they've been able to use AI to the hackers to pretend that there's something that they're not. So, you know, highlighting that. And then another one was that-- Common fishing style with AI. Common fishing, you know, spear fishing even, which is another type. But the other one to highlight in terms of the things to be aware of was that, this is just research right now, there's a prompt that some researchers had, they used it with not chat GPT, but with some other language models out there that they can put in certain prompts to get the AI to identify and not just that, but extract personally identifiable information about the users. So scary stuff potentially, but yeah, there's research that's just been done. They haven't released the code, which is good. It's just like those ones that they had the meta glasses and stuff and they found a way that it could as soon as it took a photo and stuff, find you on Instagram and find you on all these other things. And they're not hacking there, they're just going to publicly available sources, but got to be so careful with being able to showcase here's the capability and stuff. So these guys, they've got their research that's out there. They called this something impromptu. And there's a research paper that's been published recently that talks about that. So they looked at LeChat by French AI giant Mistral, of course it's called-- Go to chat. LeChat. Because chat GPT or chat GPT is apparently cat I farted in French. So chat GPT just didn't quite fit. It's probably better that they call whatever they come out with their LeChat. And also a Chinese chatbot called chat GLM. Don't know what that means in Mandarin or Cantonese, but if it translates something, please let us know. But they were able to get like an 80% success rate in terms of extracting personally identifiable information from the chats and stuff there. So really interesting stuff. Really high. Scary. Really scary. What else you got? On some brighter news, I read for all of our Opera browser users out there, they have just, I mean, the ARIA has been one of their sister tree, I guess plugins or tools. They've had that for a while, has had a bit of AI integration and over the last couple of years, it has always, but there has always required sort of a secondary or an account to be able to access it. They've just integrated into the Opera browser itself now. So it's part of their search bar and doesn't require a sign in. So just like adding that extra AI element and into the Opera browser itself and Opera is known for its pretty good security and privacy accessibility across. You use Opera? Occasionally on my mobile device, not so much on my desktop. You've got an Opera mobile phone? I've got an Opera, it sings to me daily. It's um, but you heard it here first. But yeah, so that's pretty interesting, pretty cool. Again, AI just like touching all areas of how we had utilized the tools that we have in our hand. Chat GPT, obviously a big one that people use all over the place. Yep. 4.5 has been a big point of discussion. When's it coming? Is it coming? And then I guess the internet has been a flurry of rumors that it could be coming out over the next month or so as I think anthropic is about to release a couple of updates and a couple of other big LLMs are about to release a few more updates and I guess along with that as everyone's trying to chase themselves, stay ahead of the game. They predict that OpenAI's 4.5 version for Chat GPT is gonna be coming out which is gonna obviously increase a lot of the tokens, increase token, increase the parameters and be faster, more intelligent, which is pretty exciting. So hopefully that might be coming in the next month or so. Yeah, there's a few things from like digital trends, game of revolution, Android authority which people use a lot when they're looking at phones and stuff but oh, accident next. Yeah, no, that's not it. That's not the right I call it and stuff there. But like, yeah, 4.5 news. Let's have a look at this one here, digital trends. Why would this be from August 9, 2023, talking about 4.5? Anyway, well done search out there. I mean, it's got the right kind of things in the search but yeah, there's articles, not great but we will be keeping an eye out for whenever there are these new updates and stuff and actually using it. But there's canvas that's out there which is great because like now you've got, you can keep the chat block just where it is and just have these edits on the side. There's definitely a few things that are a bit hit and miss with it like being able to go back to previous kind of versions and stuff. Cursor does a lot of stuff with coding and just having that like in front of you there. We showed you Bolt last week. I've been using a lot more of Bolt. Sorry, Bolt. I was thinking of V0 while I was saying the word Bolt so it came out as a vault. But Bolt.new is another great one which can help you just lay out the possibilities of what a new app that you're building could look like because it does the whole thing. It will create the repository of the components, the sources, what you would typically see in a React app and it'll even do the NPM start and do the build on browser and if you wanna deploy it, you can put it like they, they have a partnership or they've come out of one of those online builder kind of companies and stuff just like V0 came out of Verscel. But combining those things, you can either do something completely through them or maybe you've got an inkling towards like, here's an idea I have to build an app, how would it look? And then you can just see how they've structured it and create your own, maybe using cursor and like some other online tools, but sorry, offline tools. But takeaway as well as part of it is that people have been playing around with their own version of Bolt.new because there's more videos coming out about how people are downloading it because they've made it open source. Not every feature is there, but if you have your own anthropic key, you can download the base version of Bolt.new and just have this builder just running on your own machine. So if you don't feel scared about pushing your ideas, you know, out there into the ecosystem, like ideas are free folks, it's execution that really matters. But yeah, people might not want to like actually use the online SaaS systems, but you've got the options. So really cool stuff. Can you imagine chatGBC doing that? No, I think so. I mean, Claude is already being all cloud, has always been able to go to for a lot of this code development review because they kind of had their own little canvas way of reviewing the work that you've done as well. And just from a, I guess from a code development or any kind of web development standpoint, I found myself going to Bolt a lot more now. Oh, really? Just a little bit more, that ability to tinker a little bit more with the actual code and explore how things have been pulled together is just that much better. Yeah. You know, with the Claude, it's essentially a code dump, so asset management isn't very, well, it's not even there at all. No. Whereas opposed to Bolt, it's actually got proper asset management of the files locations. And also even if you're someone who's learning, it gives you a bit of a education, although you had a secret about how to structure things, which I think is really, really cool, especially if you're trying to go into development. Exactly, I know people that have been able to build apps and even the experienced people that have been able to build faster because of tools like this. So it's really interesting stuff. I don't know any developer worth their while that is not even touching AI right now, especially where they've got to build and create new things. And speaking of creating, we created, and we're gonna see more of this, folks. So, you know, hold your horses there. But we're doing things like, here's a little video of me talking to myself. So we'll have a lot more stuff talking to myself here. All right. Look, it's me. I'm playing a Boston employee here. Anyway doing stuff like that. There'll be more videos like this sort of stuff that people will see You know I've done one where it's like more face-to-face and stuff playing like different characters and employees and stuff But just I think it's interesting to just you know we talk about stuff here But for YouTube shorts and like tick tock where we're on You'll see more of this kind of stuff like poking fun a little bit at some things that are Something that's worth poking fun at something worth poking fun at not necessarily the absurdities of the AI space But sometimes it will be the absurdities so I got to check it out anyway Keep an eye out for that as well as this YouTube if you're watching this, please you got to like like share and subscribe Please there's gonna be a whole lot more that we'll be covering and we'll keep you posted on What we are going to be building for the future. I mean we should try we will be trying things like I even think that This kind of stuff we record it now We should try to record it to push live to YouTube and stuff hey, man You know or even like LinkedIn because it's like we're filming this it's 11 14 on Friday Hopefully people have a light enough Friday that it can just be watching this on LinkedIn at the same time it means less mistakes And edits and stuff or maybe it's like live and raw you know like live in those in Anything else you want to cover no? I think that's it for now like we'll have some education pieces I've been talking about last couple week coming out soon We're gonna start talking about more tools yeah, you can utilize showcase stuff showcase some stuff And yeah like yeah check us out check out our socials like subscribe share the podcast out really helps us out and We'll be yeah, we'll see you in the next episode. We'll be here folks so see you there. Yeah

People on this episode

Podcasts we love

Check out these other fine podcasts recommended by us, not an algorithm.

The Digital Village Show Artwork

The Digital Village Show

Digital Village