Written by: Boaz Sobrado
Compiled by: AididiaoJP, Foresight News
"I don't want to hand control of my work over to a venture capital firm," Ben Goertzel said on the "On The Margin" podcast. "Because I think AGI is too important to do that."
Goertzel popularized the term "Artificial General Intelligence" (AGI) through an influential book, led the development team for the Sophia robot, and has spent decades trying to build a thinking machine. Now, his bet goes against almost everyone else rushing toward the same goal: the most important software in human history should not be exclusively owned by a single company. "I'm a very strong believer that the core AGI code that does the thinking should be free and open source," he said.
Note: Ben Goertzel is the founder of SingularityNET, a prominent AGI researcher, and is hailed as the "AGI Godfather." He strongly advocates that AGI must be open source + decentralized, opposing monopolization by a few companies like OpenAI and Anthropic. He believes AGI is too important to be controlled by venture capital firms or a single enterprise.
This belief drives his blockchain project SingularityNET and the broader Artificial Superintelligence Alliance. The alliance was formed in 2024 by Fetch.ai, SingularityNET, and Ocean Protocol (Ocean Protocol exited by the end of 2025), merging their respective tokens into FET. While OpenAI and Anthropic raise hundreds of billions of dollars and hide their strongest models behind closed doors, Goertzel has linked his AGI work to a crypto network owned by its users.
Open Source Code Isn't Enough; Decentralized Computation Is Also Needed
His argument starts with a problem that open source alone cannot solve. He said that merely publishing code isn't very helpful if no one can afford to run it.
"If the code is open source, but the data requires a server to store it, and you need a hyperscale server to use it, then having the code open source doesn't actually help much," he said. "What you really want is to deploy the first AGI on a decentralized network spanning fifty different countries, controlled by ten thousand different people."
In his view, this is the whole point of putting AI on the blockchain. He acknowledges this might make some people nervous because bad actors could fork the system and create harmful things, but he is willing to accept this trade-off.
"I think there are more good people than bad people in the world, and they will be willing to host AI systems," Goertzel said. "We'll be better off having open and decentralized AI, rather than getting into an AGI arms race where only a few major powers have AGI and use it to destroy each other."
Those Labs That Shifted from Open Source to Closed
Goertzel is blunt about companies that once shared his commitment to openness but later strayed. He points to the court records from the lawsuit between Elon Musk and Sam Altman, which clearly show how quickly OpenAI's founding mission shifted.
"If you look at all the court hearing records between Elon Musk and Sam Altman, it seems Sam Altman didn't stick with it for very long," Goertzel said. "He shifted very quickly toward wanting to make it proprietary. The records show he pivoted quickly toward acquiring hardware, and that required them to be closed and proprietary."
He is even harsher on Anthropic's origins and equally unsparing about Musk's shift from AI doomsayer to AI builder.
"Dario Amodei's thing was closed and proprietary from the very beginning," Goertzel said. "Elon Musk's relationship with openness has always been complicated. In 2015, he first said AI is summoning the demon and nobody should build it. Then he probably started trying to build it himself. But what he's doing at xAI now is also closed AI development."
Goertzel admits the closed path is indeed simpler: raise venture capital, lock down the models, and aim for acquisition. He insists the open path is not impossible, just harder. "Linux and the internet itself are proof: you can be open source, globally decentralized, and still become a foundation that makes a lot of people money," he said.
How Token Founders Make Money
Currently, his business runs on cryptocurrency. "I have SingularityNET, which is a utility token, so it's a blockchain project," Goertzel said. "We have node operators running SingularityNET, hosting AI processes. We basically make money through the tokenomics on this network."
He expects this to change. The plan is to keep the AGI code open source while selling polished products built on top of it, with the blockchain hidden in the background. "Our project might shift slightly from Web3 toward Web2, starting to offer products you can buy with regular money," he said. "The backend is still our crypto network, but to the end user, it's just an AI service."
Goertzel says SingularityNET will launch a paid tier for businesses and power users sometime next year. "We will launch something like Claude Pro or ChatGPT Pro but smarter," he said. "It runs on a decentralized blockchain backend, but with much greater reasoning power and creativity than current chatbots." He won't chase the mass-market chatbot space. "I'm not going to try to launch a retail product like ChatGPT because those guys are all losing money," he said.
The Agent Economy He's Building
Goertzel's pitch to ordinary users is: the next wave of advantage will belong to those who can command fleets of AI Agents, not to the labs.
"The next set of wins will likely belong to small groups that can effectively organize larger teams of AI Agents to get things done," Goertzel said. "The key is how you educate and command your army of beneficial Agents."
Other builders in the crypto space are converging on the same vision: Agents that not only answer questions but also spend money. "The next step—which is already beginning—is AI Agents starting to transact on your behalf," said Varun Kabra from the identity chain Concordium on the same podcast. "They will make payments, sign up for services, and possibly handle your financial transactions."
This gives software a risky job. "The nature of an agent is outsourcing purchases, and anyone who's had experience outsourcing purchases knows there are trade-offs," said Nitya Subramanian from crypto wallet company Para on the same podcast. Goertzel's answer: run this economy on an open network, not corporate clouds.
Why This Matters
Goertzel still believes human-level AGI is coming soon. "I think we can achieve it by 2029," he said. "I wouldn't be shocked if it happens in 2027. I wouldn't be shocked if it's 2030." He is more worried about who gets left behind by the machines than what the machines themselves will do.
"If there's a gap in understanding about what's happening with AGI between the top tier and the bottom tier of the population, that gap will make inequality grow even faster," he said.
Whether a crypto network can beat companies sitting on hundreds of billions of dollars remains unproven. Goertzel's first test will launch in a few weeks. "We will launch the first downloadable version of our new Agent Omega Claw in a few weeks," he said. "We'll have the chance to teach our own personal Agents to help manage our lives and help us make money."
