Mythos, Strikes, Chip Crunch
· The Fluency Briefing
The Fluency Briefing
Your Guide to What's Happening in AI and Why It Matters to You
Wednesday, April 8, 2026
Three stories this Wednesday share an uncomfortable thread: Anthropic built an AI model so good at hacking it refused to release it publicly, ProPublica's newsroom walked off the job over AI protections, and the chips powering all of this still need a round trip to Asia before they work. The pattern?
AI's capabilities are outrunning the institutions - corporate, labor, industrial - built to contain them.
Today in AI:
- Anthropic Built a Hacker, Then Locked It Up - Anthropic announced Project Glasswing, a coalition with Apple, Google, Microsoft, and 45+ organizations to defensively deploy its unreleased Claude Mythos Preview model. Mythos found thousands of zero-day vulnerabilities, including a 27-year-old bug in OpenBSD, and can autonomously chain exploits that previous models couldn't touch. Wired
- Anthropic's Revenue Just Tripled - Anthropic's annualized revenue run rate hit $30 billion, up from $9 billion at the end of 2025, with over 1,000 business customers each spending more than $1 million annually. The company also announced a multi-gigawatt compute deal with Google and Broadcom. VentureBeat
- ProPublica Staff Strike Over AI Protections - Roughly 150 unionized employees at the nonprofit newsroom walked off the job for 24 hours on Wednesday, demanding protections around AI use, layoff safeguards, and wages as part of ongoing contract negotiations. It's reportedly the first U.S. newsroom strike centered on AI concerns. The Verge
- America's Best Chips Still Need a Passport - Even chips fabricated in the U.S. often fly to Asia for advanced packaging, and capacity is in short supply. TSMC's most advanced packaging method is growing at 80% annually, while Intel just landed Elon Musk's SpaceX, xAI, and Tesla as packaging customers. CNBC
- Google Tells Retailers: Fix Your Data or AI Can't Sell for You - Google's latest guidance warns that AI-driven shopping features like conversational search, virtual try-ons, and shoppable streaming ads only work if your product data feed is clean. Translation: garbage in, no sales out. Google Blog
- Amazon Wants AI to Host Your Podcast - Amazon launched Nova 2 Sonic, a speech-to-speech model that generates real-time conversational podcasts with two AI hosts on any topic. It supports seven languages and streams with low latency, targeting content teams that want scale without studios. AWS
- Pokemon GO's Map Data Gets a Second Life - Niantic Spatial, spun out from the Pokemon GO maker's $3.5 billion gaming sale, launched a revamped 3D mapping platform trained on billions of player-submitted images. Applications now target robotics, construction, and industrial inspection rather than catching Pikachu. GeekWire
- AI Propaganda Goes Full Lego - The Guardian reports that AI-generated content, including Lego-style animations and fake war footage, is flooding social media amid the Iran-U.S. conflict. Both sides are mixing real strikes with movie clips, video game footage, and AI fabrications, making reliable information nearly impossible to identify. The Guardian
Today's Takeaway:
Here's the thing about Anthropic's Mythos moment: it's the first time a major AI company has looked at its own creation and essentially said, "This is too dangerous for the market, so let's turn it into a public good first." Claude Mythos Preview didn't just find bugs - it autonomously wrote a browser exploit chaining four vulnerabilities together, escaped OS sandboxes, and achieved full root access on FreeBSD through a 20-gadget attack that no automated tool had caught. Previous Claude models had a near-zero success rate at this kind of autonomous exploit development. Anthropic is committing up to $100 million in usage credits and $4 million to open-source security organizations through Project Glasswing, according to Fast Company. That's defense spending, not product marketing.
But pull the lens wider and the implications get uncomfortable. This means every major software platform - your operating system, your browser, your company's servers - has vulnerabilities that a sufficiently capable AI can find and exploit. Which means the chip packaging bottleneck reported by CNBC isn't just a supply chain headache; it's a security question about where the hardware running these models gets assembled. And the ProPublica strike covered by The Verge isn't just a labor dispute - it's workers saying, "If AI capabilities are this powerful, we need contractual guardrails before management deploys them." For you, the takeaway is blunt: AI's offensive capabilities are months, not years, ahead of the defenses most organizations have in place.
💡 Fluency Moment - Building your AI fluency, one term at a time.
"AI Agent"
In plain English: An AI that takes actions on its own, completing multi-step tasks without human guidance.
Think of it like: A contractor you hire who figures out every step themselves - you just describe the finished result you want.
Why you'll hear about it: Anthropic's Mythos autonomously chains exploits, showing agents acting independently in high-stakes real-world situations.
🧰 Your Toolkit
5-Minute Quickstart: Using AI Agents to Stay on Top of the News
- Open ChatGPT or Claude in your browser and type: 'Act as my daily news assistant for [your industry or interest].'
- Ask it to summarize today's biggest stories: 'Give me 3 bullet-point summaries of important news in [AI / finance / tech] today.'
- Tell the AI your experience level: 'Explain these topics like I am completely new to [AI / investing / tech].'
- Ask a follow-up question about anything confusing: 'Why does it matter that AI chips have to go to Taiwan before they can be used?'
- Request a simple action step: 'Based on these news trends, what is one thing a beginner like me should learn about or watch this week?'
- Save your favorite AI response as a note so you can track how news topics evolve week over week.
Once you are comfortable chatting with AI about the news, try asking it to compare two stories or explain how they connect to each other. This builds a habit of thinking critically with AI as your thinking partner.
The Bottom Line
The Pattern: AI's capabilities are accelerating faster than the structures meant to govern them - whether those structures are cybersecurity defenses, labor contracts, or semiconductor supply chains. Every story today is about the gap between what AI can do and what we're prepared for it to do.
Why It Matters: If the most capable AI lab on the planet thinks its own model is too dangerous to ship, the software your business runs on has vulnerabilities you haven't imagined yet. And the people who build, package, and report on these systems are all simultaneously raising their hands to say the current rules aren't enough.
Your Move: Ask your IT team one question this week: "What's our plan if AI-discovered vulnerabilities start getting disclosed faster than we can patch?" If the answer is a blank stare, you've found your priority.
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
💬 Community | 📞 Book a Consultation | 🌐 Website
Fluently yours, The My AI Fluency Team