@MrBen

A $5 chip just verified a real Bitcoin payment entirely by itself, then moved a motor because the payment was real. No server. No cloud. No internet. Hand it a forged payment and it catches the tampering instantly and refuses. The motor stays dead. Real money moves the machine. Fake money does not. Here is how it works, in plain words. A payment of 100 satoshis arrives. The chip checks three things, all on its own silicon: that the money is real and sits inside a genuinely mined block, that it was paid to this chip specifically, and that the signature is valid. 25 milliseconds. Then the motor turns. It does all of this offline. The chip carries one number in its memory, a 32-byte fingerprint of a Bitcoin block, and proves any payment belongs to that block with a short chain of hashes. This is Simplified Payment Verification, from Satoshi's 2008 paper, Section 8, on hardware the size of a fingertip. And it barely grows. That proof stays a few hundred bytes whether the network does ten transactions a second or ten billion. The network scales up. The tiny verifier does not have to. Why it matters for AI in the real world: an AI can misread a camera, and a language model can be tricked. You cannot fully trust the judgment. But a payment cannot be faked. So the machine acts on money it can verify, not on a guess. LLM's can hallucinate. The money cannot. Two more things fall out of this. Every action leaves a public receipt of who paid for it. And you can stop the machine by simply not funding it, with no software override to hack. Check it yourself on the public BSV ledger. The block the chip carries: whatsonchain.com/block-height/9…, whose Merkle root ac9951e8... is the exact number burned into the chip. Ran 17 times on camera, on 3 separate chips, identical to the millisecond. This is one chip. Next: a swarm of small machines that pay for and verify their own actions, each leaving a receipt on a ledger no one controls. #Bitcoin #BSV #PhysicalAI #IoTs #BRC105