Bittensor (TAO)

After 20 years in traditional finance, I am going all in on decentralised AI. The AI race will shape far more than markets. It will decide who controls access to intelligence, how knowledge is produced and where power ultimately sits. So @ArtiomS and I created @xTensorCapital and @xTensorLabs. AI cannot be built entirely on infrastructure controlled by a handful of companies. We are backing the parallel path. The thesis is simple: decentralised AI remains underinvested, undervalued and widely misunderstood. It is early, technically difficult and easy to dismiss. That is exactly where asymmetry lives. xTensor Capital invests. xTensor Labs builds and operates. Each makes the other sharper. Much more to come on what we are building, learning and backing. Stay tuned!

There is literally a race for deregistration happening on Bittensor $TAO between the low priced subnets A sign that the subnet extraction strategy is no longer viable. They are being suffocated in Realtime. Bullish signal 🎯 https://t.co/MdD3bh34k8

A Bittensor subnet just built an AI safety model that beats the big players Trishool (@trishoolai), Bittensor's (@opentensor) subnet 23, released HaloGuard 1.0 today. The 4B version claims first place across seven established prompt-safety benchmarks, and the 0.8B version outperforms models several times its size. The pitch is speed over scale. HaloGuard screens prompts in real time before they reach the underlying model or agent, instead of catching harmful outputs after the fact. An earlier alpha version already runs live on the Chutes subnet, hitting an 87% F1 score on real traffic since May. The subnet's whole design is built around attacking itself. Miners get paid for breaking the model, then the breaks get patched.

A Bittensor subnet just built an AI safety model that beats the big players Trishool (@trishoolai), Bittensor's (@opentensor) subnet 23, released HaloGuard 1.0 today. The 4B version claims first place across seven established prompt-safety benchmarks, and the 0.8B version outperforms models several times its size. The pitch is speed over scale. HaloGuard screens prompts in real time before they reach the underlying model or agent, instead of catching harmful outputs after the fact. An earlier alpha version already runs live on the Chutes subnet, hitting an 87% F1 score on real traffic since May. The subnet's whole design is built around attacking itself. Miners get paid for breaking the model, then the breaks get patched.

Getting your hands on $TAO just got a whole lot easier. #AlchemyPay is proud to support @opentensor by adding $TAO to our fiat-crypto On-Ramp. Through our extensive global network, whether you use Apple Pay, Visa, or local mobile wallets, you can now power the future of decentralized AI with local fiat payments. https://t.co/IzZCQosDNu $ACH

We’re excited to announce that Trishool’s HaloGuard 1.0 𝐡𝐚𝐬 𝐚𝐜𝐡𝐢𝐞𝐯𝐞𝐝 𝐒𝐎𝐓𝐀 prompt-safety performance among open-weight guard models. Today, we present HaloGuard 1.0, a constitutional input classifier for multilingual AI safety. It is built as a first-layer input guard that checks user prompts before they reach a downstream LLM, agent, or application. This is part of the safety infrastructure being built through @trishoolai , our decentralised AI red-teaming subnet on Bittensor SN23. Full arXiv paper goes live soon.