AI: Spending, Cuts, Lawsuits
· The Fluency Briefing
The Fluency Briefing
Your Guide to What's Happening in AI and Why It Matters to You
Sunday, March 15, 2026
Three stories from this Sunday share a through-line if you squint: Meta is reportedly eyeing layoffs for 20% of its workforce to fund AI spending, the U.S. Army just handed Anduril a contract worth up to $20 billion, and ByteDance hit pause on its AI video tool after Hollywood came knocking with cease-and-desist letters. The pattern?
AI isn't just building new things - it's rearranging who gets paid, who gets cut, and who gets sued.
Today in AI:
- Anduril Lands a $20 Billion Army Deal - The U.S. Army signed a 10-year contract with Palmer Luckey's defense tech company, consolidating over 120 separate procurement actions into one enterprise deal. Translation: the Pentagon is betting big that software-defined warfare is the future. TechCrunch
- Meta Reportedly Mulling 20% Layoffs - According to Reuters, Meta is considering cuts that could affect roughly 15,800 employees to offset its massive AI infrastructure spending. Meta called the report "speculative," but the timing echoes a broader trend of companies using AI pivots to justify workforce reductions. TechCrunch
- ByteDance Freezes Global Rollout of Seedance 2.0 - After Disney and Paramount Skydance sent cease-and-desist letters over copyrighted material in its AI video generator, ByteDance reportedly suspended the tool's expansion outside China. A viral AI clip of Brad Pitt fighting Tom Cruise apparently didn't charm the studios. Engadget
- Spotify Lets You Boss Around Your Algorithm - Spotify announced Taste Profile, an AI feature that lets Premium users directly tell the algorithm what to recommend - or what to stop recommending. It's rolling out in beta in New Zealand first. Engadget
- GitHub's Slopocalypse Kills a Python Community - Jazzband, a popular open-source Python collective, is sunsetting because AI-generated spam pull requests made its open-membership model unsafe. Only 1 in 10 AI-generated PRs reportedly met project standards, and curl shut down its bug bounty after confirmation rates dropped below 5%. Simon Willison
- Zhipu AI Ships a Tiny OCR Model That Punches Up - Chinese AI lab Zhipu AI released GLM-OCR, a 0.9-billion-parameter model built to parse messy real-world documents - tables, formulas, handwriting - without requiring massive compute. If it works as advertised, small businesses could run serious document processing on modest hardware. Marktechpost
- LangChain Releases Deep Agents for Complex AI Workflows - LangChain launched Deep Agents, a structured runtime designed for multi-step AI tasks that need planning, memory, and context isolation. Standard LLM agents break down when tasks get long and stateful - this is built for that messy middle ground. Marktechpost
- Palantir's CEO Says AI Will Reshape the Electorate - Alex Karp suggested that AI-driven automation will disproportionately affect highly-educated female voters who lean Democratic, framing it as a political forecast. Whether it's a genuine warning or a thinly veiled sales pitch for Palantir's services is left as an exercise for the reader. The Guardian
Today's Takeaway:
Here's the thing about this Sunday's news: AI's biggest disruptions aren't coming from what it creates - they're coming from what it displaces. Meta reportedly considering a 20% workforce reduction to fund AI infrastructure is the starkest version of this, but the same current runs through every story. As TechCrunch notes, even OpenAI's Sam Altman has called some of these cuts "AI-washing" - executives using the AI narrative as cover for corrections that have nothing to do with automation. So what does this mean for you? If your company is spending more on AI, ask where that money is coming from.
The displacement pattern extends beyond jobs. ByteDance's Seedance 2.0 pause shows AI displacing existing intellectual property norms - the tool worked great until it started generating Brad Pitt without Brad Pitt's permission, and Engadget reports that Disney and Paramount moved fast to shut it down. Meanwhile, GitHub's open-source community is being displaced by the very AI tools developers built - Jazzband's shutdown, as documented by Simon Willison, is what happens when AI spam overwhelms human trust systems. The Anduril contract tells the same story at scale: $20 billion to replace 120 legacy procurement processes with one software-driven platform. In each case, AI isn't just adding capability. It's subtracting the old way of doing things, and nobody gave the old way a severance package.
💡 Fluency Moment - Building your AI fluency, one term at a time.
"Training Data"
In plain English: The content an AI learns from to generate outputs, like videos, text, or images.
Think of it like: Teaching a student using textbooks - but the student copied the books without asking permission first.
Why you'll hear about it: ByteDance got sued for using Hollywood films to train its AI video generator.
🧰 Your Toolkit
5-Minute Quickstart: Using AI Tools for the First Time
- Go to chat.openai.com or claude.ai and create a free account using your email address.
- Type a simple question in plain English, like 'Explain [TOPIC] as if I'm completely new to it.'
- Ask the AI to help with a real task you have today, like 'Write a short email to [PERSON] about [SUBJECT].'
- If an answer is too long or confusing, type 'Can you make that simpler and shorter?' to get a clearer reply.
- Try asking the AI to compare two things you're curious about, like 'What's the difference between [OPTION A] and [OPTION B]?'
- Save or copy any response you find useful - AI chats don't always save automatically depending on the tool you use.
Once you're comfortable chatting, try uploading a document or image to see how AI can read and summarize files for you. Exploring tools like Google Gemini or Microsoft Copilot can also show you how AI is built into apps you may already use.
The Bottom Line
The pattern across today's stories is simple: AI isn't politely joining the party - it's rearranging the furniture and handing some guests their coats. Jobs, copyright norms, open-source trust models, and defense procurement are all getting reshuffled at the same time.
Why it matters: the companies and communities that survive this transition won't be the ones that adopt AI fastest. They'll be the ones that figure out what to protect while everything shifts around them. Meta can't just slash headcount and call it a strategy. ByteDance can't ship a video tool that treats Hollywood IP like a buffet. Open-source projects can't assume good faith when bots outnumber humans.
Your move: this week, pick one process in your work that AI could automate - then ask yourself what breaks if it does. That's the question every story today forgot to answer until it was too late.
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