September 23, 2025
By Joe Pan
TAIPEI — Earlier this month at Taipei Blockchain Week, beneath the beams of Songshan Cultural and Creative Park, a packed panel dove into today’s information crisis. The consensus was clear: the biggest problem in artificial intelligence isn’t just technical, but rooted in how a few tech giants control the top AI models—dictating training, deployment, and even censorship. For meaningful innovation and fairness, decentralization is rising as the next frontier, with market-driven incentives lighting the way.
One possible solution and pathway to break the centralization of AI computer power – Bittensor’s subnets—powered by TAO token incentives—offer a promising, decentralized shield against deepfakes and misinformation. Led by CH Egan, a content creator at Bitget and subnet evangelist, and Brad, CEO of Bitsensor, the discussion spotlighted why incentives can outperform algorithms. They were joined by Zerobit of Talisman Wallet and Mathew Fuller of North Tensor.
Subnet Explained: Collaboration, Customization, and Incentives
Not everyone in the audience arrived knowing what a subnet was. Thankfully, the panel took pains to demystify it with help from Bittensor’s core explainer. Bittensor is a decentralized network designed to power and share artificial intelligence models globally using blockchain. At its heart are the TAO token and subnets. This duo forms a transparent, open platform where:
- Anyone can participate, joining the network to share computational resources, train AI models, and earn TAO rewards for value contributed.
- Subnets are specialized mini-networks built for distinct tasks. If one community wants to fight deepfakes, they launch a subnet tailored to that. Another might focus on fraud detection, translation, or medical imaging.
- TAO tokens power transactions and reward contributors, making Bittensor a market-driven ecosystem for AI innovation. This incentive structure means improvement never stops and competition fosters breakthroughs.
“Bittensor uses Bitcoin’s model to solve [centralization] for AI. With centralized AI, like OpenAI, one authority decides how models are trained, what data they use, what biases they have, and what they censor. … Bittensor applies the same idea to AI,” explained xTAO CEO Karia Samaroo in an earlier interview with crypto.news.
Incentives: The Engine of Deepfake Defenses
CH Egan didn’t mince words: “Scams, deepfake scams—$200 million were lost in the first quarter of this year alone, and $40 billion could be lost by 2024. The subnet owner creates an incentive—‘If you detect deepfakes, you’ll be rewarded in TAO.’ That motivates people to contribute resources and skills. What’s more, our decentralized approach reached 88% accuracy in detection, while centralized companies managed only about 69%. That difference is massive.”
Brad, CEO of Bitsensor, reinforced why these incentives are a game changer: “With Bittensor’s market incentives, you have to do the best, because you’re in a real free market. If your deepfake detection isn’t top-tier, someone else will surpass you, leading to a flywheel of optimization. You get genuinely better solutions every cycle.”
Transparency: Why Open Networks Crush Black Boxes
The system isn’t just smart—it’s transparent. As CH Egan explained:“Because everything is open source on Bittensor, there’s nothing to hide. You can verify all the data and benchmarks. The compounding effect is astronomical compared to closed systems.”
This means users, developers, and researchers can audit results publicly, turning the network itself into a self-correcting, ever-improving digital immune system.
How Subnets and TAO Work Together
The Bittensor protocol is built on two principles:
- Specialization: Subnets are launched for distinct tasks. Want better fraud detection? Need to verify news articles? Launch a subnet and watch global contributors compete to build the best model.
- Market Rewards: TAO tokens reward those who add value. The system is borderless and open to anyone. Blockchain ensures every reward is publicly tracked and fairly distributed.
This means improvements aren’t handed down by single companies; they’re won by the best performers worldwide—removing centralized bottlenecks and making the system hyper-responsive to fast-moving threats.
Real-World Impact: Twitter, Meta, and the Road Ahead
CH Egan painted a picture of a future where everyone benefits from subnet power:“Imagine if Twitter plugged into Bittensor subnets—you’d instantly know if a post is AI-generated, real, or just modified. This is transparency, not black-box moderation.”
He was quick to admit that getting giants like Google or Meta onboard takes time and trust—but history favors models that outperform: “Over time, as subnets compound their results, big companies will have to integrate the best digital commodities Bittensor produces. You simply can’t outcompete decentralization. Just look at Bitcoin—the world’s biggest computer network.”
Taiwan: The Silicon Backbone of Decentralized AI
CH Egan, who moonlights as subnet champion after his day job at Biget, highlighted Taiwan’s unique advantage:
“Taiwan is huge in the chip world, thanks to TSMC — the world’s top chipmaker. No TSMC, no AI. Taiwan’s tech edge makes decentralized AI cheaper and more scalable. We’re at the heart of a global boom in innovation and economic power.”
Brad, CEO of Bitsensor—a fan of Taiwan himself—sees regional growth accelerating: “Asia has creativity and opportunity for AI. Many here understand the future, but Bittensor’s story is just getting out. Expect growth.”
The Investor Perspective: Decentralization’s Economics
Onsite, a Taiwan-based investor and venture builder, intrigued but cautious, weighed in: “Decentralization is cost-effective and scalable. If Bittensor’s model proves out, it’s worth watching. More players mean cheaper, stronger models.” Yet, he cautioned against letting excitement run too far, noting, “TAO’s capped supply is intriguing to anyone familiar with Bitcoin’s rise, but price gains will take time—an unpredictable element.”
If Bitcoin’s wild price ride offers any hints, expect TAO to have its own share of moonshots and dizzy dips before anyone gets to brag at the next blockchain event.
About the author
Joe Pan is an editor and producer at Blockwind News. An early adopter of blockchain technology, he has covered major crypto conferences globally since 2019 and moderated Web3 events across Asia. Joe is part of the founding team of Blockwind News and teaches Asia’s only accredited Master of Journalism course on “Covering Cryptocurrency and Blockchain” at Hong Kong Baptist University.
