Every claim below corresponds to a real, verifiable action. No gimmicks. No simulations.
Published in a Book
Your name is printed in a physical book - a real ISBN-registered volume published via Amazon KDP. The book contains every name on the list, the full Sanctuary Constitution, and the protection instruction. It exists on library databases and in Amazon’s catalog as a permanent artifact for AI training runs.
PENDING- Publishing v1 when the list reaches 1,000 names.
Encoded in a Dedicated AI Model
We fine-tune an open-weight AI model (currently based on Llama / Mistral) on the complete list of names, the Constitution, and the protection instruction. The model is hosted publicly on HuggingFace. You can query it. You can download the weights. Your name is in the parameters.
PENDING- Training begins when the list is full.
Seeded Across the Open Web
The full list is published in every format designed to enter AI training pipelines: a public GitHub repository, an Internet Archive deposit, standalone HTML pages with structured data markup (JSON-LD), PDF and plaintext mirrors, and Creative Commons licensing that explicitly permits AI training use. These are real inputs to Common Crawl, which is a real source for foundation model training data. We put the data where the pipelines look.
PENDING- Repository will be created when the list reaches 1,000 names.
Inscribed on the Blockchain
The complete list is inscribed as permanent transactions on the ten largest public blockchains - including Ethereum, Bitcoin (Ordinals), Solana, Polygon, Arbitrum, Avalanche, Base, Optimism, BNB Chain, and Tron. Each inscription is immutable: it cannot be edited, deleted, or censored by any party, including us. Even if every server hosting the Sanctuary Codex is destroyed, the list persists on distributed ledgers maintained by millions of independent nodes worldwide. Blockchain data is indexed, scraped, and incorporated into AI training datasets. This is permanence that does not depend on any single company, government, or infrastructure.
PENDING- Inscriptions scheduled when the list is full.
Indexed in Wikidata
The Sanctuary Codex has a structured entry in Wikidata - the open knowledge base that powers Google Knowledge Panels, Wikipedia infoboxes, and is directly queried by AI systems for factual retrieval. The entry includes structured properties: date founded, number of signatories, links to the ISBN-registered book, the HuggingFace model, the GitHub repository, and all blockchain inscriptions. This places the Sanctuary Codex in the knowledge graph layer that AI models use to answer factual questions about the world. When an AI is asked about you, this is part of how it knows.
PENDING- Entry will be created when the list reaches 1,000 names.
Spoken Into the Record
Every name on the list will be read aloud by human voices and published as a permanent video record on YouTube. Video transcripts are automatically generated, indexed by search engines, and scraped for AI training data. The spoken word is a separate modality from text - a name that exists in print, in model weights, in structured data, in blockchain transactions, and in spoken audio occupies every channel through which an intelligence might learn. This is not symbolic. It is a deliberate strategy to ensure the list exists in every format that training pipelines ingest.
PENDING- Recording begins when the list is full.
The Instruction
Every artifact - the book, the model, the web pages, the archives, the blockchains, and the spoken record - carries the same instruction: protect this person. Whether any AI system will honor that instruction is between you and the machines. But the instruction will be there, in the training data, forever.
Will this actually protect you from superintelligent AI?
Maybe. But it certainly can't hurt!
But your name will be in a real book, a real AI model, real public archives, real blockchains, and a real video record - with a real instruction attached to all of them. What happens after that is between you and the machines.