AI: Fraud, Fakes, Data Harvest
· The Fluency Briefing
The Fluency Briefing
Your Guide to What's Happening in AI and Why It Matters to You
Sunday, March 22, 2026
A North Carolina man pleaded guilty to stealing millions in music royalties with AI-generated songs, a major publisher pulled a horror novel over AI suspicions, and thousands of gig workers worldwide are selling recordings of their daily lives for pennies to train AI models. Three very different stories, one uncomfortable pattern: the line between human and machine output is dissolving, and the people getting burned first are the ones who didn't see it coming.
Today in AI:
- The $10 Million Fake Playlist - Michael Smith, 52, pleaded guilty to wire fraud conspiracy after flooding streaming platforms with thousands of AI-generated songs and using bots to rack up billions of fake listens, siphoning royalties from real musicians. He faces serious federal charges in New York. The Guardian
- Your Walk to Work Is Worth $14 - Thousands of gig workers in countries like South Africa are selling videos of their feet, phone calls, and daily routines to apps like Kled AI for a few dollars a pop. The data trains AI models, but the long-term privacy implications remain murky at best. The Guardian
- Hachette Kills a Horror Novel Over AI Suspicion - Publisher Hachette pulled the horror novel 'Shy Girl' from both U.S. and U.K. markets after GoodReads reviewers and YouTube creators flagged the text as likely AI-generated. Author Mia Ballard denies using AI, blaming a hired editor. BBC
- AI Tokens as the New Signing Bonus - Nvidia CEO Jensen Huang floated the idea at GTC that engineers should receive roughly half their base salary in AI compute tokens. VC Tomasz Tunguz says startups are already doing it, putting one in five compensation dollars toward inference costs. TechCrunch
- TikTok Bans 20 Accounts Over AI-Generated Fake Influencers - The BBC found dozens of accounts using AI-generated, hypersexualized Black female avatars to funnel users toward explicit paid content. The accounts used racial tropes and weren't labeled as AI-generated. Meta says it's investigating but hasn't confirmed any action. BBC
- Your LLM Can Profile You From Your Comments - Developer Simon Willison demonstrated that feeding 1,000 Hacker News comments into Claude produces a startlingly accurate personal profile, including professional identity, political leanings, and personality traits. The data is publicly available through Algolia's API. Simon Willison
- ChatGPT Makes Learning Easier and More Persuasive - A new study found that AI overviews and chatbot-style answers don't just help people learn faster - they're also more persuasive than traditional search results. That's a feature if the information is accurate and a real problem if it isn't. TechXplore
Today's Takeaway:
Here's the thing about this Sunday's news: nearly every story is about someone trying to pass off machine output as human work - or human data as worthless raw material. Michael Smith flooded Spotify and Apple Music with AI songs and bot listens, extracting millions that should have gone to actual artists. Hachette pulled 'Shy Girl' because readers - not the publisher - spotted the telltale signs of AI-generated prose. On TikTok, fake AI influencers exploited racial stereotypes to drive traffic to paid sites, with no disclosure labels in sight. According to The Guardian, gig workers are selling intimate slices of their lives for $14 a clip to train the very models powering these deceptions.
Through an economic lens, the pattern is clear: AI has made it trivially cheap to produce convincing fakes but painfully expensive to detect them. Streaming platforms couldn't catch Smith's fraud for years. Hachette's own editorial process missed the AI text until internet sleuths flagged it. And the gig workers feeding data into these models have no visibility into what their footage ultimately trains. The cost of creating synthetic content is plummeting toward zero, but the cost of verifying authenticity - in music, publishing, social media, and identity - is climbing fast. That gap is where the fraud, exploitation, and ethical mess lives, and it's only getting wider.
💡 Fluency Moment - Building your AI fluency, one term at a time.
"Training Data"
In plain English: The real-world examples fed to an AI so it learns how to behave.
Think of it like: Like showing a child thousands of flashcards so they learn to recognize cats, words, or faces.
Why you'll hear about it: Gig workers are literally selling their daily lives to become AI training data.
🧰 Your Toolkit
Try This Prompt: Understanding Today's Biggest AI Stories
Explain to me like I'm a curious teenager: how can someone take a real video of a person and use AI to change what they look like or do in it? What makes this dangerous, and what should I watch out for online? I've heard about satellites, but what exactly is 'Low Earth Orbit' and why are big companies suddenly spending billions to put things up there? Give me a simple explanation using everyday comparisons I'd actually understand. Some tech companies are starting to pay workers with 'AI tokens' instead of just money. Explain what AI tokens are, why a company might offer them, and whether [YOUR JOB OR INDUSTRY] workers might ever see something like this. I want to understand how AI can read thousands of social media comments and build a profile of who someone is. Walk me through how this works using a simple everyday example, and tell me what this means for my privacy on platforms like [PLATFORM YOU USE, e.g. Reddit or Twitter]. Summarize the most important thing I, as someone new to AI, should understand about [CHOOSE A TOPIC: deepfake videos / AI in space / AI as payment / AI profiling people online] - and give me one simple action I can take this week to stay informed or protected.
For the best results, paste these prompts into ChatGPT or Google Gemini and add your own details in the brackets - the more specific you are, the more useful and personal the answer will be.
The Bottom Line
The Pattern: Every story this Sunday points to the same fracture - it's now cheaper to generate than to verify. AI can produce songs, novels, influencer personas, and personal profiles at near-zero cost, but spotting the fakes still requires human vigilance, investigative journalism, or a mob of skeptical GoodReads reviewers.
Why It Matters: If you run a business, create content, or simply consume media, the burden of proof is shifting onto you. Platforms aren't catching this stuff fast enough, publishers are missing it entirely, and the economic incentives for fraud are only growing as AI tools get better and cheaper.
Your Move: Pick one thing you consumed this week - a song, an article, a social media post - and spend two minutes asking whether a human actually made it. That discomfort you feel is the new literacy.
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