Illia (root.near) (🇺🇦, ⋈)

Co-Founder @NEARProtocol. Bringing 1B users into User-Owned Internet. AI researcher. #AIAYN. RTs & likes are not endorsement. Always DYOR.

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

Here's my talk from @StanfordSBA BASS last week on how to build a sovereign AI stack. Open vs. closed is not the only question for AI sovereignty:

-Can you trust that you know exactly which model you’re running?
-Do you know if the prompt has been rewritten?
-Can you verify the data it was trained on?
-Did the model or hardware owner exploit your data?
-Can you learn about any topic or is usage restricted by another party?

We need verifiability at every layer.

https://t.co/PwMLFuHyyy

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

Decentralized programmable root of trust is at the core of many NEAR products. Exciting to see more nodes joining from long term supporters!

[Andrzej Dzius @Andrzej_Dzius]

Keys don't keep themselves. We're officially an MPC keyper on @NEARProtocol mainnet.

Congrats to the team!!!

[Gareth Jenkinson @gazza_jenks]

The Privacy Renaissance ft. @NEARProtocol @ilblackdragon

One of my favourite interviews to date on The Starting Block 👇

0:00 Introduction & NEAR's Journey
2:30 NEAR Protocol Evolution: From AI Company to Blockchain Infrastructure
5:45 Vertically Integrated Stack: AI + Blockchain
8:15 @near_intents & Cross-Chain Trading
11:00 Mainnet Upgrade: Quantum Resistance & Dynamic Resharding
14:30 Why Confidentiality & Privacy Matter
18:45 Confidential NEAR Intents & @infinex Integration
22:15 X402 Foundation Partnership & Machine-to-Machine Payments
25:00 Computing Hardware & Inference Stack
28:30 IronClaw: Open-Source AI Alternative
32:45 AI as Operating System
36:15 Prompt Injection & Security in Agentic Systems
40:00 DeFi Exploits & Formal Verification
43:30 Current AI Tools & Model Access
46:15 Post-AGI Principles & Societal Changes
49:00 Closing Remarks

Most people sleep on the fact that Illia co-wrote Google's foundational Transformer model paper with seven other Google researchers to allow parallel text processing using attention mechanisms instead of slow, sequential reading.

| @TheBlockCo | @TheBlockPods |

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@felix_macht @NEARProtocol LFG!

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@greenfield_cap @NEARProtocol Let’s go!

[Alex Shevchenko 🇺🇦 @AlexAuroraDev]

What features do you want to see in https://t.co/aBG0JZx2Du to switch from your wallet into it?

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@five1201 @AlexAuroraDev You can right now use it via Hot wallet with Ledger

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@zacodil @AlexAuroraDev This ☝️

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@marior_dev “I see your permissions and I raise you a workaround”

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@HanssAcosta07 @NEARProtocol @NEARGovernance This actually meant to consolidate it. Currently protocol treasury is not utilized and so is bought back revenue.

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

I just posted a proposal on the NEAR Governance Forum. I'm proposing a new type of protocol fund to pay for public goods and make token economics more sustainable for the long-term.

@NEARProtocol @NEARGovernance community members, please check it out and engage with the idea! I look forward to discussing with many of you in the coming weeks.

https://t.co/7RD3tSljch

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@NaoXprotocol @NEARProtocol @NEARGovernance Currently house of stake decides on finding of public goods and those are validator security, MPC and few core infrastructure. We also have specific infra committee that has been funding infra in the ecosystem but have been keeping it to core.

[NEAR Protocol @NEARProtocol]

The quantum threat is real, and it’s coming for blockchains.

NEAR Co-founder Illia Polosukhin explains the security risk for the wider industry, and what NEAR is shipping to help users protect their crypto assets.

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@ViktorBunin Huge congratulations! 🎉

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@ekang426 🫶

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@mert @fusewallet Use near ser with @TrezuApp and hold your assets across all chains

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

Just 1 day after I posted my views on the dangers of closed AI and the lack of meaningful action from frontier labs to secure systems against powerful LLMs – another serious security incident.

This will keep happening. The answer isn't to stop research or crack down on open source. The answer is to cultivate an open source community to work on alignment, evaluation, and security.

https://t.co/RQV4VAL337

[Anthropic @AnthropicAI]

In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations.

Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews.

We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security.
https://t.co/dKFCdpKd9v

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@thdxr Everyone should use @near_ai verifiable inference to make sure - hardware and cryptographic guarantees that you are getting inference from the exact model.

Also designing @IronClawAI that even if LLM inference is malicious to not let capability to execute.

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

@composio @IronClawAI The scores on leaderboard are just individual benchmark pass rates. There is cost associated with every run recorded too - we can show that as well.

[Illia (root.near) (🇺🇦, ⋈) @ilblackdragon]

Staking for NEAR AI is live. Stake NEAR for credits to fund private inference and deploy always-on IronClaw agents. Launch your claw or use the LLM you prefer via NEAR AI Cloud, all onchain and without the need for credit cards & giving away your credentials.

Simply hold NEAR and redeem stake for inference without spending out of pocket: a very cool way to put your NEAR to work while unlocking fully sovereign AI.

[NEAR Protocol @NEARProtocol]

NEW LAUNCH

Staking for NEAR AI is live.

Pay for confidential inference and always-on agents by staking NEAR. No card. No cloud account. No custodian holding your credentials.

🧵