bpaceIND🏕️ (@buildspaceIND) header
bpaceIND🏕️ (@buildspaceIND) profile picture

building

Joined January 2022
Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)Photo by bpaceIND🏕️ (@buildspaceIND)
bpaceIND🏕️ retweeted
Been seeing a lot of takes on how to protect against prompt injection. It's a structural problem that can't be solved with filters or better prompting. The same thing that makes LLMs useful makes injection possible. Instead, focus on negating the damage. Agents need strong isolation, limited access to secrets, and enough observability to know when an injection occurred. And when one does, blow up the agent's environment and start fresh.
1
5
12
3,765
AI is progressing rapidly: GPT-5.4 Pro (xhigh) has achieved a massive 10 point gain in CritPt, a benchmark where the highest score was only 9% in Nov ‘25 This is the largest incremental gain we have seen from a single release. CritPt is a benchmark with a private dataset that tests performance on research-level physics reasoning tasks. When CritPt was released in November 2025 the highest score was 9% (Gemini 3 Pro Preview). Only ~4 months later the highest score has more than tripled to 30%.
32
116
22
1,010
113,872
bpaceIND🏕️ retweeted
New on the Anthropic Engineering Blog: In evaluating Claude Opus 4.6 on BrowseComp, we found cases where the model recognized the test, then found and decrypted answers to it—raising questions about eval integrity in web-enabled environments. Read more: anthropic.com/engineering/ev…
237
345
143
3,108
1,086,165
bpaceIND🏕️ retweeted
DeFi will win.
343
244
103
2,355
182,966
bpaceIND🏕️ retweeted
Twin goals of EF: 1. Usage of Ethereum - but usage of a type where users are actually benefiting from Ethereum's underlying properties 2. Resilience and decentralization of Ethereum - viewed holistically, focused on addressing points of weakness anywhere in the stack
Replying to @VitalikButerin
@VitalikButerin and I wrote this to share why the EF exists and what’s next—real-world usage & deeper resilience. With @hwwonx & @tkstanczak co-leading EF’s next chapter, we're also counting on the community to reach those goals
523
395
94
2,426
923,501
bpaceIND🏕️ retweeted
By the way, the recent @VitalikButerin proposal around replacing the EVM with RISC-V was based on results we found in this paper about the 800x provability overhead of EVM in SP1. Some numbers from the paper:
Introducing vApps: Verifiable Applications at Internet Scale vApp is a Rust-based development framework for applications to build on crypto rails with internet-grade performance. When implemented with a zkVM, benchmark results show consistently high performance: • Up to 832× fewer cycles than EVM interpreters • 95.7% reduction in proving cycles via precompiles • 30× throughput gain with GPU acceleration • 346MB → 1.5MB proof compression via recursion By @LayerZero_Core and @SuccinctLabs, the vApp framework is the optimal model for building applications that use blockchains at internet scale. Develop like web2, with the verifiability of web3.
265
248
36
951
255,003
bpaceIND🏕️ retweeted
lagrange's DeepProve lays the groundwork for trustless AI using zkps, it brings zkML to life in Turing Roulette game (turing.lagrange.dev) where we can guess if we're chatting with a human or an AI & has generated over 200k proofs already it's backed by a prover network AVS secured via @eigencloud that ensures every AI interaction is verifiable, tamper-proof & private breakdown of @lagrangedev’s zkML architecture to understand how -
2
3
11
784
In a mostly open source industry like crypto, the only thing that really matters is speed of execution. Ideas are free. Expecting more and more vertical integration. Being an excellent, but ultimately, thin layer of the stack no longer sufficient.
26
32
17
302
37,038
bpaceIND🏕️ retweeted
Coded ZKP/FHE, a decentralized, collaborative, robust ZKP/FHE System By: - 0x1cc 🔗 ethresear.ch/t/19459 Highlights: - The research introduces a method to break down complex ZKP/FHE tasks into smaller subtasks, distributing these across multiple nodes for parallel processing. -Redundancy is added to the subtasks, meaning the original task's result can be reconstructed from a subset of the subtask results, enhancing the system's robustness against delays or failures. -A (n,k) coded ZKP/FHE system is proposed, where a task is divided into n subtasks, and the completion of any k subtasks is sufficient to obtain the original task's result. -This coded design significantly improves computation speed within ZKP/FHE systems and allows the system to tolerate the downtime or delay of n-k nodes, making it more robust. -The coded ZKP/FHE system facilitates the creation of decentralized, collaborative, and resilient cryptographic operations, promising enhanced efficiency and reliability for applications like zkRollup and fheEVM computation. ELI5: Imagine you have a big puzzle to solve, but it's really hard to do it all by yourself. So, you decide to break the puzzle into smaller pieces and ask your friends to help you solve each piece. Some of your friends might be slow or even forget to solve their piece. To make sure you can still see the whole picture without waiting forever, you give out some extra pieces (redundancy) that can help you figure out the missing parts if some friends don't come through. This way, as long as you get enough pieces back, you can still see the whole picture. This is similar to what we're doing with complex math problems (ZKP/FHE tasks) in computers, making the process faster, more reliable, and allowing everyone to help out from their own computers.
14
1
59
6,963
bpaceIND🏕️ retweeted
Built from Mumbai, for the world 🧡
11
18
12
160
13,619
bpaceIND🏕️ retweeted
"One bad signature will be able to drain your account on Ethereum after EIP-3074." Yes; this is true. 3074 coauthor here! Let me put this concern to rest a bit before it gets more out of hand.
32
251
42
1,019
223,126
bpaceIND🏕️ retweeted
Fraud Proofs Are Broken By: - GCdePaula 🔗 ethresear.ch/t/19234 Highlights: - Optimistic rollups rely on fraud proofs to ensure security, but current algorithms are vulnerable to Sybil attacks, affecting safety, speed, and decentralization. -Three properties for analyzing permissionless interactive fraud proofs are safety, promptness, and decentralization, but achieving all three simultaneously is challenging. -Permissionless Refereed Tournaments (PRT), Bounded Liquidity Delay (BoLD), and OP Stack’s Fault Proof System (OPFP) are three algorithms aiming for permissionless interactive fraud proofs, each with its own strengths and weaknesses. -The 'proof of whale' attack is a significant threat to the safety of fraud proof systems, where a well-funded adversary can overwhelm honest validators by sheer financial force. -Collaboration and open research are essential for improving fraud proof algorithms and ensuring the security and decentralization of optimistic rollups. ELI5: Imagine you're playing a game where everyone has to agree on the rules and the outcome of the game. But some players might try to cheat to win. In the world of online transactions, 'fraud proofs' are like a system that lets any honest player prove when someone is trying to cheat, ensuring that the game remains fair for everyone. However, this article explains that the current ways we try to catch cheaters have some weaknesses, making it possible for them to still get away with cheating in certain situations.
2
4
2
42
3,077
Blobs are live on Ethereum Mainnet! The Dencun fork finalized without issues and the network has been running stably since then. We can see healthy blob propagation statistics (expect a deeper analysis over the next weeks). Congrats to all the client teams and contributors!
4
22
6
167
16,231
I'm excited to share our latest research on adjusting the block gas limit. Together with @VitalikButerin, we're exploring paths to reduce the maximum block size, balance calldata pricing, and ultimately raise the block gas limit. ethresear.ch/t/on-increasing…
5
73
7
252
35,017
bpaceIND🏕️ retweeted
The best term I can think of for satisfying the "clearly differentiate from IPFS" criterion is "bulletin board". The term even has a not-just-blockchain-world heritage! eprint.iacr.org/2021/047.pdf
14
9
1
70
40,828
bpaceIND🏕️ retweeted
By popular demand, an updated roadmap diagram for 2023!
1,605
3,797
1,112
15,617
3,713,301