Writing

Will AI agents hack your business? From high-profile AI lab incidents to the world’s first fully autonomous ransomware (JadePuffer), digital threats have fundamentally evolved. Yet, no matter how sophisticated these attacks become, they still rely on the same old weakness:

Our human tendency to prioritize convenience over safety.

In this new article, I'm breaking down the rising risk of AI-driven attacks and the quiet insurance gaps that leave small business owners exposed. Then, I will show you how to defend your business using simple security guardrails.

The core principle is straightforward: you just need to make your business a much harder target so you can use digital tools with total confidence. For this, you protect your business by adding intentional friction where the risk is highest.

While AI agents can execute complex attacks in seconds, they have much more trouble when they hit stops. By mastering two basic security fundamentals, securing access (authentication) and isolating key operations (keeping them separate), you build speedbumps that trap small mistakes before they become costly emergencies.

𝘛𝘩𝘢𝘯𝘬𝘴 𝘵𝘰 𝘑𝘦𝘯𝘯𝘪𝘧𝘦𝘳 𝘚𝘮𝘢𝘳𝘵 𝘧𝘰𝘳 𝘵𝘩𝘦 𝘪𝘥𝘦𝘢 𝘰𝘧 𝘵𝘩𝘪𝘴 𝘢𝘳𝘵𝘪𝘤𝘭𝘦.

𝘛𝘩𝘢𝘯𝘬𝘴 𝘵𝘰 𝘙𝘪𝘤𝘢𝘳𝘥𝘰 𝘥𝘦 𝘖𝘭𝘪𝘷𝘦𝘪𝘳𝘢 𝘧𝘰𝘳 𝘩𝘪𝘴 𝘳𝘦𝘷𝘪𝘦𝘸 𝘢𝘯𝘥 𝘤𝘰𝘮𝘮𝘦𝘯𝘵𝘴. Article Link (LinkedIn).

The video clip went viral in late July. A staffer printed the draft without checking it, and the politician read it without catching the mistake. It’s funny, but before we point fingers: 𝗶𝘁 𝗰𝗼𝘂𝗹𝗱 𝗵𝗮𝗽𝗽𝗲𝗻 𝘁𝗼 𝗮𝗻𝘆 𝗼𝗳 𝘂𝘀.

I can easily picture myself. I pitching a big product to a new client and I click to the next slide. And then we both watch a chatbot prompt pop up on screen.

I’d turn bright red and want the floor to swallow me whole. But why? Mistakes happen. And sometimes, the consequences are far worse than an awkward meeting. Look at the lawyers submitting fake court cases to judges, or the driver who steered into a frozen lake because a GPS told them to turn.

𝗪𝗵𝘆 𝗱𝗼 𝘄𝗲 𝘁𝗿𝘂𝘀𝘁 𝘁𝗵𝗲𝘀𝗲 𝗼𝘂𝘁𝗽𝘂𝘁𝘀 𝘀𝗼 𝗯𝗹𝗶𝗻𝗱𝗹𝘆?

It turns out our brains love to save energy. When we see bad grammar or messy formatting, a mental alarm goes off. We slow down and double-check. But when we see total nonsense written smoothly, the brain assumes it’s accurate. No alarm goes off. We just hit copy and paste.

Willpower won't fix this. We need simple, practical rules at work (like mandatory review steps) before we make our own embarrassing slipups.

Have you ever almost trusted an AI answer just because it sounded well-written?

Check out my latest article to explore more high-profile AI mistakes, the psychology behind why we fall for them, and how teams can build simple guardrails.

𝘛𝘩𝘢𝘯𝘬𝘴 𝘵𝘰 𝘙𝘪𝘤𝘢𝘳𝘥𝘰 𝘥𝘦 𝘖𝘭𝘪𝘷𝘦𝘪𝘳𝘢 𝘧𝘰𝘳 𝘵𝘩𝘦 𝘰𝘳𝘪𝘨𝘪𝘯𝘢𝘭 𝘪𝘥𝘦𝘢 𝘰𝘧 𝘵𝘩𝘪𝘴 𝘢𝘳𝘵𝘪𝘤𝘭𝘦, 𝘩𝘪𝘴 𝘳𝘦𝘷𝘪𝘦𝘸, 𝘢𝘯𝘥 𝘩𝘪𝘴 𝘪𝘯𝘴𝘪𝘨𝘩𝘵𝘴. Article Link (LinkedIn).

This law does not just mean adding "powered by AI" disclaimers to your chatbots and goes much deeper.

𝗛𝗼𝘄 𝗔𝗿𝘁𝗶𝗰𝗹𝗲 𝟱𝟬 𝗖𝗵𝗮𝗻𝗴𝗲𝘀 𝘁𝗵𝗲 𝗥𝘂𝗹𝗲𝘀 𝗳𝗼𝗿 𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀𝗲𝘀:

- 𝗛𝗶𝗱𝗱𝗲𝗻 𝗗𝗶𝗴𝗶𝘁𝗮𝗹 𝗧𝗮𝗴𝘀: Software makers must build invisible, permanent tags directly into AI-made files. If an employee creates a product image or audio clip using AI, the file carries a hidden digital tag that platforms and search engines read automatically.

- 𝗖𝗼𝗺𝗽𝗮𝗻𝘆 𝗕𝗹𝗮𝗺𝗲 𝗳𝗼𝗿 𝗗𝗲𝗲𝗽𝗳𝗮𝗸𝗲𝘀: If an employee posts a realistic fake video of an executive or an altered event photo on official company accounts without a clear badge, the company takes the blame and faces heavy fines.

- 𝗥𝘂𝗹𝗲𝘀 𝗳𝗼𝗿 𝗔𝗜 𝗡𝗲𝘄𝘀 𝘃𝘀. 𝗠𝗮𝗿𝗸𝗲𝘁𝗶𝗻𝗴: Standard AI marketing emails do not need labels. But if a company uses AI to write articles meant to inform the public about news, politics, or money, it must show a clear label—unless a human editor checks the facts and takes full responsibility for the story.

- 𝗡𝗼𝘁𝗶𝗳𝘆𝗶𝗻𝗴 𝗣𝗲𝗼𝗽𝗹𝗲 𝗕𝗲𝗳𝗼𝗿𝗲 𝗦𝗰𝗮𝗻𝘀: If a company uses AI to check voice tones, expressions, or physical traits (like tracking customer frustration on support calls), it must warn people before the check starts.

𝗧𝗵𝗲 𝗚𝗹𝗼𝗯𝗮𝗹 𝗥𝗶𝗽𝗽𝗹𝗲 𝗘𝗳𝗳𝗲𝗰𝘁:

For global companies, building separate software versions just for Europe costs too much money. To stay compliant in Europe, these companies are building these mandatory disclosure tools into the software everyone uses daily.

I broke down the practical business impact and key rules in my latest deep-dive article. Article Link (LinkedIn).

Your hands are covered in grease. Your muscles are burning holding an assembly. And this phone starts ringing on top of a step ladder three feet away. You watch the screen flash. But if you let go, steel drops on your chest. You miss the call, the customer hangs up, and... you guess they'll dial your competitor.

That missed call is money handed straight to someone else.

I wrote a story about this headache, and I shared how to solve it yourself using basic and low-cost AI tools.

Instead of paying a software vendor $200 every month, you can connect a few simple pay-as-you-go services together for around $15.

When you can’t answer, the AI picks up in a professional voice. It can ask if it's an emergency and grab the customer's address. And then, it texts a short summary straight to your wristwatch. You keep working without ruining your phone screen or losing the job.

This isn't a sales pitch, and I'm not selling software. It’s about 𝘁𝗮𝗸𝗶𝗻𝗴 𝗼𝘄𝗻𝗲𝗿𝘀𝗵𝗶𝗽 𝗼𝗳 𝘆𝗼𝘂𝗿 𝗼𝘄𝗻 𝘁𝗲𝗰𝗵.

In fact, the technical barrier is low. Does your son play video games or tweaks basic app settings? They can follow the guide and have this running in an afternoon. No coding required.

Read the full story and step-by-step setup guide. Article Link (LinkedIn).

Artificial Intelligence is everywhere in office tools, promised as a way to boost productivity and keep us focused on core business. But in reality, relying on more tech to communicate has flooded our inboxes with information while making authentic 𝗵𝘂𝗺𝗮𝗻 𝗰𝗼𝗻𝗻𝗲𝗰𝘁𝗶𝗼𝗻 rarer.

When collaboration is delegated to software rather than direct and personal communication, cognitive fatigue sets in. Managers then end up spending valuable hours untangling miscommunications and subtle friction that didn't exist before, all of which quietly drags down 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘃𝗶𝘁𝘆.

I'm very optimistic about AI's potential in business. However, I believe we need to design and use AI to foster real 𝗰𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝗼𝗻 and protect employee well-being through clear 𝗯𝗼𝘂𝗻𝗱𝗮𝗿𝗶𝗲𝘀.

In my new article, I break down:

- Why "performative AI empathy" backfires (and turns off your team)

- How AI can translate heavy jargon into real, human stories

- How automated gatekeeping lowers turnover, prevents burnout, and keeps projects on schedule.

Stop letting tech push your team backward. Read how to build a thoughtful and communication-first AI strategy today. Article Link (LinkedIn).

Imagine deploying autonomous software to balance electricity supply and demand in real time. If your code glitches, you could trigger a regional blackout. Because autonomous software shifts risk from physical hardware to digital logic, traditional insurers can't easily underwrite it without ironclad proof.

That is where 𝗱𝘂𝗮𝗹 𝘃𝗮𝗹𝗶𝗱𝗮𝘁𝗶𝗼𝗻 comes in.

In this article, I explore why 𝗪𝗲𝘀𝘁𝗲𝗿𝗻 𝗖𝗮𝗻𝗮𝗱𝗮 is a world-class proving ground for grid-scale AI. Because of extreme climate, vast distances, and contrasting regulatory models all at the same place.

I break down:

- What insurers need: specific safety records, real-world data, and emergency backup plans.

- The Dual-Province Strategy: Why testing a technology across two completely different systems, like Alberta’s open market and BC’s government utility, proves it can succeed elsewhere in the world.

- Two Practical Examples: Case studies on smart hardware for instant repairs and software for neighborhood energy trading, showing how to lower risk for investors.

By validating AI solutions through the regulatory and environmental gauntlet of Western Canada, innovators build an insurable and export-ready product. Article Link (LinkedIn).

Imagine your business has built its reporting workflow around one AI service. Then the provider changes its pricing, removes a feature, or changes its terms. Your data is still yours, but replacing that AI means redesigning workflows, retraining staff, and investing time and money.

That's what AI sovereignty looks like in practice.

In this article, I explore what Canada's new "𝗔𝗜 𝗳𝗼𝗿 𝗔𝗹𝗹" strategy means for Alberta businesses. Beyond access to AI, I look at how businesses can keep control over their data, avoid unnecessary vendor lock-in, and preserve the freedom to adapt as AI evolves.

I also propose several practical ideas that could help Alberta businesses.. Article Link (LinkedIn).

When you prompt an agent, you delegate the reasoning layer to a system that matches semantic patterns rather than business context. If a prompt misaligns, a calculator can output incorrect margins, leading to financial loss.

Furthermore, security is invisible during development. By default, agents optimize for a working interface, not threat modeling. A functional database built without data validation leaves sensitive data exposed to theft.

The breakthrough of agentic computing requires realism. Coding agents work when humans shift from manual writing to strategic verification.

I wrote an article breaking down why human oversight determines the success of AI development. Article Link (LinkedIn).