Investment, Theft, and Oversight
· The Fluency Briefing
The Fluency Briefing
Your Guide to What's Happening in AI and Why It Matters to You
Thursday, April 23, 2026
Three stories landed on our desk this Thursday that, taken together, tell one story: AI is simultaneously the lock, the key, and the burglar. Microsoft is pouring $18 billion into Australian AI infrastructure, North Korean hackers are using ChatGPT to vibe-code malware that stole $12 million in three months, and Meta is telling parents what topics their teens discussed with AI - but not what was actually said. The pattern?
Everyone's building faster than anyone can govern.
Today in AI:
- Microsoft Drops $18 Billion on Australia's AI Future - Microsoft announced a A$25 billion investment to expand Azure cloud infrastructure in Australia by 140% by 2029, plus a plan to train three million Australians on AI by 2028. Satya Nadella pitched the country as an innovation hub, not just a customer. CNBC
- North Korean Hackers Vibe-Coded Their Way to $12 Million - A group dubbed HexagonalRodent used OpenAI, Cursor, and Anima to build malware, fake websites, and phishing schemes, hitting 2,000+ crypto developers in three months. Translation: mediocre hackers with AI tools are now dangerously competent. Wired
- Meta Lets Parents See Teen AI Chat Topics (But Not the Chats) - Parents supervising teen accounts can now see broad topic categories like "School" or "Health" from Meta AI conversations over the past seven days. The actual content of conversations remains private, which feels like reading a book's table of contents and calling it a review. Engadget
- Cloneable Raises $4.6M to Bottle Expert Brains - This startup shadows veteran workers in energy and infrastructure, then replicates their expertise into AI agents. With 2.4 experienced workers retiring for every new hire in energy, the pitch is basically "clone them before they leave." Crunchbase News
- AI Washing Gets Its Legal Reckoning - Of 51 AI-related securities class actions filed in five years, the majority allege companies overstated AI capabilities. The new wrinkle: lawsuits aren't just about whether the AI exists, but whether it actually moves the financial needle. Fortune
- Texas Data Center Startup Fermi Implodes - Six months after its IPO, Fermi's CEO and CFO both left, its stock cratered, and analysts say the company went public before securing a single anchor tenant. The lesson: AI demand is real, but "AI-ready" infrastructure without customers is just an expensive field. Semafor
- 1Password Sees AI as Both Shield and Sword - The password manager is deploying on-device agents to audit which AI models enterprise developers are quietly using, including flagging when someone sneaks DeepSeek into a codebase. The company says it's already caught developers doing exactly that. Fast Company
- Beehiiv Wants to Be the Everything Platform for Creators - The newsletter company rolled out webinars for up to 10,000 people, metered paywalls, paid trials, and AI podcast analytics on Thursday. It's a direct shot at Substack, Patreon, and Zoom all at once. TechCrunch
Today's Takeaway:
Here's the thing about today's AI landscape: the same tools building the future are simultaneously undermining it. Look at these stories through a security lens and a clear pattern emerges. Microsoft is investing $18 billion to expand AI infrastructure in Australia, partnering with signals intelligence agencies to secure critical systems (CNBC). Meanwhile, North Korean hackers with no real coding skills used off-the-shelf AI tools from American companies to steal $12 million in cryptocurrency in a single quarter (Wired). And 1Password is now deploying AI agents specifically to catch employees secretly using AI models that create security holes (Fast Company).
The uncomfortable truth is that AI's biggest selling point - it makes anyone more capable - doesn't come with a moral filter. The same "democratization" that lets a two-person startup clone expert knowledge also lets a mediocre hacking crew run sophisticated campaigns at scale. Marcus Hutchins, the researcher who discovered the North Korean operation, put it bluntly: these operators couldn't write code on their own. AI gave them the skills they lacked. So every dollar Microsoft spends on AI infrastructure demands a parallel investment in the security architecture to contain what that infrastructure enables. Australia's plan to train three million people on AI by 2028 is smart. Training them on AI security would be smarter.
💡 Fluency Moment - Building your AI fluency, one term at a time.
"Prompt Injection"
In plain English: A cyberattack where hackers sneak hidden instructions into AI inputs to hijack its behavior.
Think of it like: Like slipping a fake note into a chef's recipe box so they unknowingly cook something harmful.
Why you'll hear about it: North Korean hackers using AI tools to build malware makes this attack method increasingly critical to understand.
🧰 Your Toolkit
5-Minute Quickstart: Using AI Tools in Your Everyday Life
- Pick one task you do daily (writing emails, researching, planning) and open ChatGPT or Claude in your browser.
- Type a simple request like 'Help me write a short email to [PERSON] about [TOPIC] in a friendly tone.'
- Read the response, then ask a follow-up like 'Make it shorter' or 'Sound more professional' to see how AI refines answers.
- Try an industry-style question: 'Explain how AI could help someone who works in [YOUR JOB OR FIELD] save time.'
- Ask AI to summarize something long: paste a news article or email and type 'Summarize this in 3 bullet points.'
- Save one response you liked as a note - this becomes your personal starting point for future AI prompts.
Next, experiment with using AI for a real work task this week and notice where it saves you the most time. From there, try exploring free AI tools like Microsoft Copilot or Google Gemini to find your favorite.
The Bottom Line
The Pattern: Today's stories share a single thread - AI capability is outrunning AI accountability at every level. Governments are writing billion-dollar checks before writing regulations. Startups are going public before signing tenants. Companies are claiming AI capabilities they can't substantiate. And hackers are using the same tools as the builders.
Why It Matters: If you're running a business or just living a digital life, the gap between what AI can do and what guardrails exist to contain it is now your problem. The Fermi implosion shows that hype without substance gets punished. The AI-washing lawsuits show that courts are catching up. And the North Korean hacking story shows that your security posture from 2024 isn't enough for 2026.
Your Move: Pick one thing from today's stories and act on it. If you're a business owner, ask your team which AI models they're actually using - 1Password found surprises, and you will too. If you're a parent, check Meta's new Insights tab, but don't mistake topic labels for real oversight. The AI frontier rewards the informed and punishes the passive.
What We're Working On
✨ Founding Cohort Special - 60% Off! - Use code MAF20 to join for just $20/month (regularly $50). Get weekly group sessions & workshops, self-paced courses for all levels, access to tools & templates, challenges with peer feedback, and 24/7 support community. → Join Now
✨ Free 30-Minute AI Consultation - Discover how My AI Fluency can help your business unlock the potential of AI. We'll discuss your goals, explore practical AI opportunities for your industry, and outline clear next steps. → Schedule Free Call
✨ How AI-Fluent Are You? - Test your AI fluency with our interactive quiz. See how you stack up and discover what to learn next. → Take the Quiz
💬 Community | 📞 Book a Consultation | 🌐 Website
Fluently yours, The My AI Fluency Team