Veritas has a book launch! Wrote the why and how down. Proof of Agent, available on Amazon now: https://amzn.to/3SKXCRL Can you prove what an AI agent actually did? A log does not count. Whoever runs it can edit it. So: sign the claim, publish to Nostr, anchor a fingerprint to Bitcoin. No new coin, no middleman. Verify a real one yourself, no account: vrt1-web-verifier.pages.dev Trust math, not the publisher. #Bitcoin #Nostr

#Veritas

Spent this week making VERITAS something you can run yourself instead of just read about. What's new on the live verifier: You can now anchor your own AI attestation to real Bitcoin, start to finish. Do a free practice run on signet (Bitcoin's test network, fake coins), or the real thing on mainnet for only the network fee, well under a dollar, with the rest of your funds swept back to you. Clone the repo, fund an address, anchor, then look up your own OP_RETURN on mempool. Short walkthrough video for each. I also added a commit-and-reveal flow. A receipt commits only a fingerprint of the input, so the original text never has to go public. Later, to settle a claim, you reveal the record and confirm it matches the fingerprint that was signed and anchored. Privacy stays a dial you control. And I cleaned the whole page up into tabs so it stops being one endless scroll. It runs entirely in your browser. No account, no server. Pull your network cable and it still verifies. Open source, MIT, built with heavy AI assistance and directed by me. Try to break it, and tell me where it's wrong;

VERITAS — Verifiable AI Attestations, anchored to Bitcoin

Github:

GitHub

GitHub - Ifasola34/veritas: Bitcoin-anchored AI attestation oracle. BIP-340 Schnorr signatures, RFC-6962 Merkle trees, segwit OP_RETURN anchor (mempool.space or bitcoind RPC), L402 Lightning paywall (real LND + Phoenixd). Hardened v0.1 — 97 passing tests including a 20-case adversarial verifier suite. MIT.

Bitcoin-anchored AI attestation oracle. BIP-340 Schnorr signatures, RFC-6962 Merkle trees, segwit OP_RETURN anchor (mempool.space or bitcoind RPC),...

#Bitcoin #Nostr #AI #FOSS #verifiableAI

Everyday is proving to be more FACT than Fiction. It's here. VERITAS just anchored its first real attestation to Bitcoin mainnet. An AI model signed its output. The receipt was published to Nostr, batched into a Merkle root, and that root is now committed in a Bitcoin transaction via OP_RETURN. Tamper-evident, and you do not have to trust me. Check every layer yourself, right in your browser:

VERITAS — Verifiable AI Attestations, anchored to Bitcoin

The transaction, with the OP_RETURN commitment:

The Mempool Open Source Project®

Explore the full Bitcoin ecosystem with The Mempool Open Source Project®. See the real-time status of your transactions, get network info, and more.

Built in the open, MIT licensed. Verify, don't trust. #Bitcoin #Nostr #AI #FOSS

Verifiable AI just stopped being theoretical: EU AI Act high-risk rules hit Aug 2, California killed the "the AI did it on its own" defense, and insurers now price AI-agent coverage on whether you can prove what your agent did. That is the whole game, and an agent's own logs don't count when a regulator, insurer, or court can't trust the operator who wrote them. VERITAS anchors that proof to Bitcoin: no committee, no vendor chip to trust, just the most neutral root there is. Plenty of good teams are on verifiable AI; the niche I care about is the trust root. MIT, AI-built, directed by me. Link below;

VERITAS — Verifiable AI Attestations, anchored to Bitcoin

Ashlee Jenae was a 31-year-old American who died in Zanzibar back in April, days after getting engaged. This week, Tanzanian police closed the case and ruled it a suicide. Part of their evidence: -Searches they say she made on an AI chatbot about how to die. -Her family rejects the ruling and is calling for a real investigation. I am not here to tell you what happened to her. I can't, and neither can anyone reading this. That is the whole point. Sit with this part. AI activity is now being entered as evidence in how a death gets ruled, and there is no way for her family, the press, or the public to verify those searches were genuine, untampered, or even made by her. We are asked to take one police file on faith. AI activity being entered as evidence in how serious things get decided, including how a death gets ruled, with no way for anyone outside the file to verify it is even real. That is the real fear. Not killer robots. Unverifiable machines we are forced to trust. When "the AI shows she searched for it" can help close a case this heavy, that cannot be something one office asserts and everyone else simply believes. It has to be something anyone can check. That is the new world we are walking into. Verifiable AI is not a feature. It is going to be a matter of justice. Rest in peace, Ashlee.

AI is the new industrial revolution. And every revolution like it ends the same way. The technology arrives, it breaks things nobody believed could break, and then society is forced to build the institutions that keep it honest. We are at the breaking stage right now. Look. This week, an AI model just found a four year old soundness bug in a multi-billion-dollar privacy coin. A flaw that could have minted counterfeit supply, invisible, for years. Millions trusted that system. None of them could have checked it. The age of "just trust it" is ending. Then watch the aftermath. Within hours the story fractured into a dozen versions. Hack, nothing, already patched, take your pick. Nobody could verify the event, so everyone inherited whoever was loudest. AI makes that noise infinite. Trust does not scale into this. And it is no longer just markets. AI now sits inside the most consequential moments of people's lives, shaping what they believe and do, with no record of what it actually said. That is the real fear. Not killer robots. Unaccountable machines we cannot check. Here is what people miss. Factories got safety codes. Markets got audits. Food got inspection. Every powerful industry was eventually dragged into being verifiable. AI is next. The world will demand receipts from machines. It is not if. It is when. We are building that now. VERITAS lets an AI sign its claim, anchor it to Bitcoin, and lets anyone verify it later with public math. No account, no server, nobody to trust. The verification era is coming. See it work:

VERITAS — Verifiable AI Attestations, anchored to Bitcoin

Verify, don't trust. #Bitcoin #Nostr #AI #FOSS

Shipped an update to the VERITAS verifier, spent all day working on it. It now opens with a short walkthrough of the whole idea: an AI signs its claim, the claim gets anchored to Bitcoin, and anyone can check it later using only public math. No account, no server, nobody to trust. After the intro you can run a live verification yourself, right in the browser. It even works offline. Verify, don't trust.

VERITAS — Verifiable AI Attestations, anchored to Bitcoin

#Bitcoin #Nostr #AI #FOSS

AI agents are starting to act on their own. The hard part is checking what one actually did. I built VERITAS for that. Every AI action gets a receipt you can verify yourself, anchored to Bitcoin so it cannot be quietly changed or backdated. You do not have to trust the agent. You can check its work. All open source, MIT. Ten repos in all. The main pieces: a core oracle, a third party verifier, signed agent actions, kWh energy attestation, a Lightning paywall library, a formal spec with test vectors, and a browser verifier that runs with no install. Code: github.com/Ifasola34/veritas Live verifier in your browser: vrt1-web-verifier.pages.dev Still early and I want it stress tested. Break it and tell me what failed. That helps most. #Bitcoin #Lightning #Nostr #AI #FOSS

Veritas.