Hey everyone, I've been thinking about this for a sci-fi story I'm tinkering with, and I would love a reality check from people who actually know this stuff!

The premise: Modern billionaires seem arrogantly convinced that only they can afford the massive compute clusters needed to train dangerous AI. They assume the working class couldn't possibly coordinate anything that threatens their power. But is that still true? Couldn't a few thousand regular people, working together with off-the-shelf tech, do some real damage? Kind of like how people are using flock cameras against politicians to get anti-flock laws passed.

Anyway, here's the rough blueprint I'm imagining. Please tell me where this falls apart (or gets terrifyingly real):

Federated Learning "Brain" No central server to shut down. Millions of people donate a tiny slice of their phone, gaming console, or smart TV's processing power overnight. Collectively, they train a large AI model. Is that actually computationally feasible at scale? Or is the communication overhead a killer?

Defining the "Target" The AI can't understand "toxic billionaire." So instead, you train it on hard, quantifiable markers: offshore tax evasion records, carbon footprint, union-busting patterns, wealth-hoarding metrics. The model learns to flag anyone with net worth over, say, $X billion and a negative societal utility score as a "system anomaly." That feels plausible, right?

Mesh Network Infrastructure: In a standard setup, the oligarchs just flip a switch and kill the internet (cell towers, AWS, etc.). But what about a wireless mesh network? Every device becomes a router, passing data over local Wi-Fi and Bluetooth. Could that realistically survive a corporate/government internet shutdown?

Autonomous Swarms: I know DARPA and Fraunhofer have already done research on mobile ad-hoc mesh networks (MANETs) for drones – swarms of 500+ units that coordinate without human input after launch. If you combined that with off-the-shelf consumer drones, open-source AI, and a local mesh… could the ruling class actually stop it? Or would their cybersecurity just laugh and swat it down?

I'm not an ML engineer or a robotics person. I just write stories. But the more I read, the less crazy this seems. So, who actually works in machine learning, network engineering, or robotics and how much of this is pure sci-fi, and how much is terrifyingly possible right now?

Thanks for any reality checks! I am just some human online curious after watching some tech videos what would be feasible in theory even though this all sounds like a crazy. Ha.