Hakr.RBT @dwebchap

The first two known exploits against live ZK circuits just happened, and they weren't subtle underconstrained bugs. They were Groth16 verifiers deployed without completing the trusted setup ceremony. One was white-hat rescued for ~$1.5M, the other drained for 5 ETH. 🧵

Web3 scale champions always compared them to Visa/ MC. The reality is V&MC dont have s/w tps.They grow horizontally on demand. Traditional monolthic designs be it in web2/web3 will always have a software limit. what web3 need is horizontal scale - only provided by @RubixChain

KC @KCRubix

5 months ago

Authentication is the real bottleneck for Agentic payments - p2p is the solution. Centralized auth gateways process ~ 100 bn logins a day. #Visa processes 674 mn txns per day. Even if agents do 100 mn tps (#Stripe estimates upto 1 bn), need 8.64 trillion auth requests!

1 3 4 339 1

3. This is where @Agent_DNA becomes cruical for Agentic commerce.It cannot depend on central auth with large p99 latency. Auth must be p2p not dependent on global RT. @Agent_DNA replaces login bottleneck with distributed trust graph using underlying @RubixChain tech

2. 0.8s avg is not the only problem.It's also p99 latency. Humans rarely reach p99 stage because their behavior creates organic load distribution.They dont retry right away.Agents behave differently. It operate continuously&retry instantly meaning p99 is inevitable

1. Typical SSO redirect based login takes 0.5s to 2 seconds . If your model follow this, The math is At 100M actions/sec with 0.8s validation time: Validations needed = requests PS × time = 100M×0.8 = 80M concurrent auth req No centralized IAM cluster can handle that.

KC @KCRubix

5 months ago

Authentication is the real bottleneck for Agentic payments - p2p is the solution. Centralized auth gateways process ~ 100 bn logins a day. #Visa processes 674 mn txns per day. Even if agents do 100 mn tps (#Stripe estimates upto 1 bn), need 8.64 trillion auth requests!

1 3 4 339 1

Today systems can verify identity far easily than they can verify intent.2025 @Cloudflare ban @perplexity_ai having same root.Current security designs blur line btn user intent & extraction.Without intent aware governance,security measures default to only ratelimits & heuristics

Anthropic @AnthropicAI

5 months ago

These attacks are growing in intensity and sophistication. Addressing them will require rapid, coordinated action among industry players, policymakers, and the broader AI community. Read more: anthropic.com/news/detecting…

353 351 7K 2.4M 1K

Using @Agent_DNA , agents working together to mount large scale scraping or distillation will be traceable and potentially prevented because their identities and delegation chains are visible.

Anthropic @AnthropicAI

5 months ago

We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax. These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.

7K 6K 54K 33.9M 13K

🆕 Brave researchers have uncovered vulnerabilities in zkLogin, a widely-deployed authorization system for blockchain transactions. Our findings demonstrate the wider challenges facing zero-knowledge proof systems.

@IBM says these 5 should be in your strategy for Agent Identity & Governance. Obeservability (logs) has to be made tamperproof to have security relevance. Ephemeral and context based access to resources and unique ID with reputation. @Agent_DNA

"Do something that brings something actually new to the table " . How about flipping the table? L1s , L2s , appchains , the problem is same. Monolithic design - shared space. @RubixChain is the only DLT protocol that doesnt talk about tps. Because tps is a hardware problem

vitalik.eth @VitalikButerin

6 months ago

Have been following reactions to what I said about L2s about 1.5 days ago. One important thing that I believe is: "make yet another EVM chain and add an optimistic bridge to Ethereum with a 1 week delay" is to infra what forking Compound is to governance - something we've done

2K 630 5K 1.1M 920

@KCRubix @RubixChain @VitalikButerin Chain by definition is linear whereas graphs allows you to branch and merge , perfect for complex decision making and error correction that modern AI systems require.

Sharded chains then ? Thats not new. Scalable ZKEVM is still years away and will always be inferior to new gen DLTs. You need a protocol that scales & settles horizontally. @RubixChain microservices of web3. Where there is no software tps.

vitalik.eth @VitalikButerin

6 months ago

There have recently been some discussions on the ongoing role of L2s in the Ethereum ecosystem, especially in the face of two facts: * L2s' progress to stage 2 (and, secondarily, on interop) has been far slower and more difficult than originally expected * L1 itself is scaling,

3K 2K 10K 6.4M 3K

We are deploying AI agents with huge access but no clear identity or accountability. They are not users, not services just powerful "things". Identity-first approaches (like what @Agent_DNA explores) feel less like a feature and more like a necessity

We need an urgent shift from Identity as Possession (Static Bearer Tokens) to Identity as State (Attestation + Runtime Context + User Intent) @Agent_DNA .

@thevdm_ @Agent_DNA is logging causality - the chain of events and interactions. Not the internal state or memory. This requires cryptographic rails on a chain like @RubixChain that scales horizontally in real time.

A log you can alter provides zero security. AI agents run with elevated privileges ,meaning they can rewrite their own history. At that point, “logging” is theater. Forensic integrity requires cryptographic certainty, not procedural comfort.

KC @KCRubix

6 months ago

Been saying this for a while. AI Agents will upend security. We need new models for security, based on decentralization, immutable logs and run time data access. @Agent_DNA leverages @RubixChain graphs to deliver AI security at scale. fierce-network.com/cloud/spontane…

1 2 2 111 0

Blockchain validators have same problem as bank lockers . It protects the “box” but not the value inside the box. There is no financial guarantee on stored value. In @RubixChain , validators not only execute state but underwrite correctness using economic collateral.

Auth tokens are session based. But user to agent delegations are not session based, they are query based. Passing Auth tokens to agents enables replay, misuse and shadow actions. Autonomous agents need intent bound authorization, not reusable session keys. @Agent_DNA

@RubixChain is scaling "state" , not trust. The same finality model , same data availability assurance , same consensus. Sharded state and sharded execution in a unified trust model.

Hakr.RBT @dwebchap

9 months ago

Rollup centric scaling assumes L2s inherit Ethereum security but in practice each L2 has different proving systems. L2 - L2 moves introduces new trust assumptions and hence you can no longer call Ethereum + L2s as a unified blockchain.

1 1 2 364 0

A unified blockchain requires: 1. one proving system 2. one DA layer 3. one governance model 4. one censorship model 5. one finality Model and many more...

The interop layer solutions only gives shared ordering , standardized messaging, smoother communication. It makes UX better but does not solve the underlying architecture problems. L2s will still, by design differ in key mgmt , data availability, censorship guarantees and more

Rollup centric scaling assumes L2s inherit Ethereum security but in practice each L2 has different proving systems. L2 - L2 moves introduces new trust assumptions and hence you can no longer call Ethereum + L2s as a unified blockchain.