DaZZee I.T.

There’s a subtle change coming to online scams. And it’s not the kind you’re watching out for 👀 When generative AI first arrived, there was a lot of talk about dynamic websites. Pages that wouldn’t be built once and shown to everyone, but generated on the fly, shaped by your location, device, behavior, even what you typed to get there. That future never really showed up. But it turns out someone’s very interested in it. Security researchers have been exploring how this idea could be used in phishing attacks, and the results are uncomfortable at best 😬 Let me explain… You click a link and land on a webpage that looks harmless. There’s no obvious malware. Nothing suspicious for security tools to grab hold of. But once the page loads, it asks a legitimate AI service to generate code in real time. That code is then assembled and run directly in your browser. The outcome is a fully working phishing page created especially for your visit.
 Different code each time. No fixed “bad page” to analyze. Nothing obvious moving across the network. Which makes traditional detection much harder. To reassure you, this is mostly proof-of-concept right now. The researchers didn’t say they’ve seen this exact technique used live yet. But they were clear that all the pieces already exist. AI is already being used to write heavily disguised JavaScript.
 AI-assisted malware and ransomware are increasing fast. 
 Dynamic code execution on compromised machines is already common. Put that together and dynamically generated phishing pages start to feel less like science fiction and more like a preview. The conclusion is that this is where scams are heading. Detection will still be possible, but it will rely more on behavior and context, not just spotting a known “bad” website. They also flagged tighter controls around which AI tools are allowed at work, and stronger security in AI platforms themselves. The bigger shift here is psychological. We’re used to thinking “that page looks fake”. But what happens when the page looks different every time? 🤔 Consider this: If scams stop being static and start being personalized, what will you rely on to decide what’s real and what isn’t?

Microsoft has taken another big step in the AI arms race 🤖 This time it’s with Maia 200. A brand-new AI chip designed and built by Microsoft itself. Now, before your eyes glaze over at the word chip 😴 I promise, this matters to everyday businesses… AI tools don’t just exist in the cloud. They run on real, physical hardware inside data centers. The faster and more efficient that hardware is, the better AI tools perform. And the cheaper they are to run at scale. Maia 200 is the next generation of Microsoft’s own AI hardware. It’s purpose-built for AI workloads, meaning it can run very large AI models using fewer machines, less power, and less wasted effort. Simply put: More work done, with less kit 💪 This also explains why Microsoft is doing it. By designing its own AI chips, Microsoft can make Microsoft Azure a faster and more efficient place to run AI than rivals, like Amazon Web Services and Google Cloud. Whoever controls the hardware gets a big say in performance, pricing, and reliability. And this isn’t a future promise. Microsoft is already using Maia 200 to power parts of Microsoft 365 Copilot and its internal AI platforms. It’s rolling out across US data centers first, with more regions to follow, and developers and researchers are being invited to test it early. You don’t need to understand the technical specs to spot the pattern 🚀 AI is shifting from a clever feature to foundational infrastructure, like electricity, internet, or cloud computing before it. The businesses building that infrastructure now are shaping how powerful, affordable, and dependable AI becomes for everyone else. So, here’s the question I’ll leave you with 👇 When AI becomes as ordinary as email in business, do you want to be playing catch up or ahead of your competitors?

OneNote getting messy? Here’s a simple way to highlight what matters so you can find it again fast…

Here’s a counter-intuitive AI tip I didn’t expect to be sharing 🤖 Being mean to ChatGPT can sometimes get you better answers. Before you ask, no, I’m not having a bad day 🤣 Tools like ChatGPT and Microsoft Copilot are what’s called generative AI. That means they don’t look up answers like Google. They generate replies based on patterns they’ve seen before. Sometimes that goes brilliantly. Sometimes they confidently make things up. That’s why you’ll often see the little warning at the bottom saying it can make mistakes. Even Sam Altman, the CEO of OpenAI, has said he’s surprised by how much people trust ChatGPT, given that it can “hallucinate” (AI-speak for confidently being wrong). But get this… Researchers at Pennsylvania State University ran a study using an older version of ChatGPT. They asked it the same questions in different ways. Polite prompts. Neutral prompts. And rude ones. The rude ones performed better 😡 Noticeably better. Short, blunt, even mildly insulting instructions produced more accurate answers than overly polite, flowery requests. The theory is that direct language reduces ambiguity. The AI focuses on the task, not the tone. Before you unleash your inner Gordon Ramsey 😅 the researchers were clear this isn’t a free pass to be unpleasant. Normalizing rude language has downsides. And future AI models may simply ignore tone altogether. The real skill isn’t being nice or nasty. It’s being clear. If you say: “Can you maybe help me understand this, if that’s okay?”, you’ll often get a vague answer back. If you say: “Explain this in simple terms. Assume I’m not technical. Give me a practical example.” the quality jumps immediately. That’s prompt engineering (aka “learning how to ask better questions”). Both Microsoft and OpenAI have said most AI frustrations come down to poor prompts, not bad technology. There’s also growing evidence that leaning on AI too heavily can dull critical thinking and confidence over time. So, it shouldn’t replace your judgment, just support it. So no, don’t be cruel to AI. But do be firm. Clear. Specific. And a little less polite if politeness is getting in the way of precision 🙂 Have you noticed a difference when you change how you ask AI for help? 👀

Digital fraud isn’t on the rise. It’s evolving. Fast. Scammers are using smarter tools, more convincing messages and pressure tactics designed to make even careful people slip up. These are the simple habits that could stop your team from falling for them…

What to Do If You Think You Clicked on a Phishing Link https://hubs.li/Q0491Zzm0

Not all tools are created equal. Some quietly support how your business works. Others add friction, risk, and lost time. It’s time to take a closer look at the difference between tools that are fit for the way you work and those that hold your team back…

Important: Protect your business from digital fraud https://hubs.li/Q048RRVH0

AI adoption isn’t splitting businesses into “tech-savvy” and “anti-tech”. It’s splitting them into those moving at different speeds inside the same company 😬 New research shows a big age-related gap in how people use AI at work. Roughly half of under-35s are already using AI tools regularly. Many have had training. Most see AI as helpful for their jobs. But around half of over-45s haven’t used AI at all. Not because they don’t trust it or because they think it’s dangerous.

 Mainly because it feels unfamiliar. And that’s where the real risk sits ⚠️ 
 When adoption is uneven, AI goes underground. Some staff quietly use AI to move faster. Others avoid it completely.
 Managers assume “we’re not really using AI yet”, when really, parts of the business already are. That creates problems like inconsistent outputs, unclear data handling and no shared standards. 
 And of course, no confidence about what information is being fed into which tools. The research also highlights something important: Countries and organizations with slower, more cautious adoption aren’t falling behind because of a lack of tools. They’re falling behind because of a lack of confidence and guidance 🤷♂️ AI doesn’t need to be everywhere to be useful. But it does need to be understood. The businesses that get the most value won’t be the ones chasing every new AI feature. They’ll be the ones that:
 • Set clear boundaries
 • Give people simple, practical training
 • And focus on using AI to remove friction, not create anxiety ❓ Is AI in your business something you’ve consciously decided how to use, or is it being used quietly, inconsistently, and without a plan?

There’s a lot of noise about AI right now, but this caught my eye because it’s refreshingly honest 🙂 A report shows that around 70% of retailers are already testing or partially using agentic AI. But only 8% have rolled it out fully across their business. In other words, most people are experimenting. Very few have cracked it. And I’m certain it doesn’t only apply to retail. 
 Agentic AI isn’t just a chatbot answering questions. It’s AI that can look across systems, spot issues, and suggest (or trigger) actions. Think delays, bottlenecks, stock problems, or inefficiencies. Not marketing slogans. Retailers are optimistic. Nearly all believe AI will be essential to staying competitive, and many expect efficiency gains very soon. But they’re also hitting reality. The biggest blockers? • Data that isn’t clean or joined up • Concerns about trust, transparency, and regulation • And a shortage of people who know how to implement AI properly What’s interesting is where AI is heading. So far, most use has been in customer service and marketing. But the next wave is about operations. Things like inventory, supply chains, fulfilment, admin. Less creative AI, more quietly fixing problems before customers notice. And that’s the bit business owners should pay attention to. The real value of AI is removing friction from day-to-day operations and freeing humans to focus on decisions that need judgment. AI works best when the foundations are solid: Good data, clear processes, and realistic expectations. So, here’s my question for you 🤔 If AI could spot problems in your operations before they became issues, would your systems be ready to support it?