History Protocol helps developers survive AI attacks with ‘programmable IP’ crypto

Jason Zhao was working as a product manager at Google's cutting-edge AI division, DeepMind, when he realized the problem of generating AI by mixing existing content into something new.

It has since been proven correct. AI is now reading news articles for free through services like Perplexity, which threatens journalism's business model. AI was also used to skip the voices of artists like Drake and The Weeknd's viral track “Heart on My Sleeve.”

The track came within a whisker of a Grammy nomination before Recording Academy CEO Harvey Mason Jr. realized that the music industry couldn't survive if neither the artists nor the label were getting paid.

“I think we're at a point where you can't ignore these questions because, what is the business model of The New York Times?” Zhao, founder of History Protocol, asked online from Palo Alto.

“You have these amazing creative tools, but if you're an IP owner, or a platform, or an individual creator, you don't have any kind of sustainable business model.”

Zhao personally thought Drake and The Weeknd's track was “really cool,” a Harry Potter AI-generated viral video shot by luxury fashion house Balenciaga, but noted that neither had a sustainable business model. For intellectual property owners or creators who use material without permission.

“JK Rowling doesn't win, the person who created the video and the community doesn't win, and so you have this lose/lose, opposite situation. The New York Times Sues OpenAI.

“And that's where blockchains come in and history comes in.”

Zhao said copyright laws were created in the 15th century, the day after Gutenberg's printing press, and could not keep up with changes in technology.

“We're still basically using this pen-and-paper legal system to figure out how to distribute IP and monetize social media, and now, with AI creating new content, mixing things up like never before, it's not going to work.” He says.

“The AI ​​model doesn't work for companies. It does not work for IP catalogs. It doesn't work for anyone.

What is a history protocol?

The historical protocol answer is to make IP programmable, in the same way that Ethereum made money programmable. It's an open market for IP, allowing rights holders to record their songs, patents or writings — or even Pink Floyd's back catalog — and then work with a contract and payment program that allows others to use it. Those fees are paid via cryptocurrency, but targeting regularity, creators and rights holders can be paid in their local currency.

“It's really seamless for the developer. You can accept payments and credit cards and pay creators in fiat. So, you can think of the blockchain layer as an abstraction.

Its high-profile backers include Polychain Capital, Hashed, Paris Hilton 11:11 Media and Samsung Next, a16z is the lead investor in a $29.3 million seed round in May 2023 and a $25 million Series A in September 2023. As of August 2024, Story has raised $80 million more in the series.

Also Read: 5 Surprising Use Cases for Based Agents and the Near AI Assistant

The relationship with a16z became even more important when US President Donald Trump named VC firm Siram Krishna as his top AI policy advisor. Krishnan wrote a long essay for the New York Times last year calling for a decentralized restoration of democracy to the Internet and a “fundamentally different way for websites to exchange value with AI assistants.”

Currently, in beta testing on the Ethereum Sepolia test network, the History Protocol mainnet is scheduled to launch in the first quarter of 2025.

Why does OpenAI agree to pay IP creators?

AI model creators are currently cleaning up some of the best online content for free, so why would you want to license and pay for IP?

Zhao suggests that they already are in some cases. OpenAI has signed deals with Reddit for $60 million and with The Wall Street Journal and The Times publisher News Corp for $250 million. It also has an extensive agreement with Stock Photo Library Shutterstock.

“Our thesis is that ultimately, it's very important for these model companies to have their own channels and to be able to access IP in a clean and simple way. The last thing you want to think about is, I'm going to be sued out of existence?”

Zhao likened the historical protocol's ability to simplify IP licensing to the way Spotify and Netflix made legal music and TV so cheap and convenient that most illegal downloading stopped.

“Spotify has solved that problem by making it so convenient to use the world's IP library that it doesn't make sense for anyone to go and try to steal it unless they have a specific reason,” he explained.

“It's the same with history. We're not saying, ‘Hey, we've built software that makes sure no one can copy IP from someone else if they don't use history.' But really, if you're OpenAI, we've created a system where you can access a very large repository of IP with a lot of different sources, and all the terms are very transparent. [set out]So you don't need to worry about it. It is more convenient for you to do the right thing. “

Empower people to avoid the failure of the AI ​​model

Another compelling reason why AI model companies want to pay for IP is to ensure that their training data does not dry up.

Research has shown that AI models trained on the output of other AI models gradually begin to produce worse and worse content through a process called “model failure.” So, while AI companies may be able to get away with scraping content for a while, they will eventually need to encourage the creation of new content.

Also read: AI agents are a hot topic in crypto trading, but beware of rookie mistakes

“Web2's social contract is that if I'm a travel blogger, if it's decent, I'll get some eyeballs, then I can run some ads and make some money,” he explains.

But someone asked the perplexity, “What should I do in Seoul?” If he asks, he will point. And the travel blog is reorganized as part of the response, then no one clicks to the blog, and no one advertises.

“And now I have no reason to create this blog because no one is looking at it. I can't monetize it. I have no reason to continue,” he says.

“In the long run, you're going to have the tragedy of collective communities taking and taking IP from this open data field, but there's no incentive to replace it. And I think in the long run, if you're producing valuable content and IP, there's going to be some kind of win-win/way to monetize that.” There must be a win-win framework.

How did Jason Zhao discover blockchain and AI?

The 25-year-old entrepreneur played around with Dogecoin mining years ago and bought a few ICOs in 2017, but “didn't take blockchain seriously until DeFi summer” in 2020.

A philosophy and computer science student at Stanford, he was fascinated by how different political philosophies could be mapped and lived in crypt, from the libertarian individualism of Bitcoin to the collaborative governance of DAOs.

At DeepMind, crypto's lightning-fast timescales were another appeal, compared to the glacially slow progress of turning research into useful products. “A white paper is basically researched, then two days or two months later it goes into production, and now about 10 million people are using it,” he said. “So that was really fun for me.”

He wasn't interested in starting another Diffie project, so he and co-founder Seung Yoon “SY” Lee started thinking about how to build a blockchain that doesn't focus on money.

“We wanted to think about the ways in which blockchain could become a ledger of assets and create a market for non-cash things. And that's where we started thinking around IP in this direction.

How does it work under the hood?

“IP Assets” are registered with immutable tokens, but these use the updated ERC-6551 Ethereum standard, which turns NFTs into smart accounts to control other assets, make transactions and interact with smart contracts and DApps.

This allows Story Protocol to automate licensing and allows users to add upfront fees, royalty percentages, and usage limits.

A Programmable IP License (PIL) links the legal contract to the smart contract, making the rights legally enforceable and enforceable onchain. Complies with the Berne Convention, which applies to most markets worldwide.

It is realizing Lawrence Lessig's vision of turning code into law and law into code.

It sounds extremely technical – give me an example.

Zhao gives the example of an artist creating a character and documenting it in history. An app built on top of the protocol could give users the ability to drag and drop that character into a comic.

“So now, your character is being used in the comic book. If that ridiculous micropayments are enabled, you'll be entitled to a share of the royalties or revenue you've set as per your terms.

One of the 50 or so applications built on the protocol so far is called Magma and has 2.5 million users. It allows groups of up to 50 to create and collaborate on digital art simultaneously.

“Artists can pull from properties across history, including those registered in Magma, and use them in their own art and track when those IPs have been used.”

Identify who is paid for what when multiple IP assets are used. Character creators can set their own terms or let apps or AI decide on a reasonable fee. This is something Zhao is happy to let the market decide.

“We're building the tools for developers to build applications around IP, and so at our product level, at our protocol, we're clearly not concerned with how to price and value IP.”

Leaning towards AI agents

Being the hottest narrative of the year, Story Protocol leans towards AI. Last week, AI unveiled the Agent Control Protocol for IP (ATCP/IP), which enables agents to negotiate complex agreements and contracts over IP.

Earlier this year, it announced a partnership with Ritual to post user-created AI models to the protocol to track text, image and audio outputs from the models. Mychel, a platform, allows users to build AI apps and characters, host them in virtual reality, and record them in a story.

Another big benefit for Stories could be content verification, so users can easily verify that an article is from The New York Times or AI-generated bullshit pretending to be true.

An identity verification system is being developed, Zhao said.

“This is an important part of the metadata associated with each ID,” he says. It's something we hope to have before we go online.

Andrew Fenton

Based in Melbourne, Andrew Fenton is a journalist and editor covering cryptocurrency and blockchain. He has worked as a film journalist for News Corp Australia, SA Wind and national entertainment writer for Melbourne Weekly.

Follow the author @andrewfenton