Co-Founder @MetaTrustLabs

Joined January 2023
The launch of OpenAI’s Operator represents a pivotal moment in the development of autonomous AI agents. This evolution signals a critical shift toward more interactive, task-oriented AI agents that could transform workflows across industries. As AI agents evolve, their ability to handle increasingly complex environments will expand, unlocking new opportunities for automation and efficiency. #AIAgents
A research preview of Operator, an agent that can use its own browser to perform tasks for you.
13
5
19
8,045
✨New achievement of Metatrust Labs: Meet VERITE - A Revolutionizing Smart Contract Security with Profitable Vulnerability Detection Smart contracts now manage billions in assets, making them lucrative targets for exploitation. However, existing fuzzing tools often fall short in identifying vulnerabilities with financial profit potential and automating their exploitation. Enter VERITE, a cutting-edge fuzzing framework that bridges this critical gap. What Makes VERITE Stand Out? 🔍 Action-Based Mutators VERITE simulates real-world actions (like token swaps and transfers) to explore fund flow scenarios, enabling the discovery of vulnerabilities rooted in financial operations. 💰 Profitability Recognition It can identify potentially lucrative exploits by analyzing token flows and uncovering anomalies, such as imbalanced token pairs or unconditional token gains. 📈 Profit Maximization Using advanced gradient descent techniques, VERITE refines exploit parameters, maximizing the profit potential from identified vulnerabilities. Key Performance Highlights: Tested on 61 real-world DeFi projects (average loss: $1.1M per project), VERITE uncovered vulnerabilities yielding an astonishing $18M in profits. Outperformed top fuzzers like ItyFuzz, discovering more profitable exploits in 12 targets and identifying 6 new vulnerabilities during audits. Applications: 1️⃣ Security Audits VERITE helps auditors detect critical vulnerabilities, elevating security assessments to new heights. 2️⃣ On-Chain Defense The framework can proactively monitor contracts and mitigate risks by identifying and exploiting vulnerabilities before malicious actors do. metatrust.io/blogs/post/intr… #Blockchain #SmartContracts #Security #DeFi
1
3
1
4
727
Calling on Researchers: Nominate a Research Paper for Competition and Win a Prize exp.science/media/calling-re… Do you know of a pioneering research paper that has contributed to breakthroughs in Distributed Ledger Technology (DLT), Artificial Intelligence (AI), or IoT?
1
609
😱LLMs show significant biases during role-playing🔥 🚨Check out our recent paper “Benchmarking Bias in Large Language Models during Role-Playing,” where we uncover biases in GPT-4o-mini, DeepSeek, Qwen, Llama, and Mistral. 📑arxiv.org/abs/2411.00585 📊github.com/LLMBias/BiasLens
1
4
1
15
2,465
Professor Meta retweeted
🎉 Thrilled by the incredible enthusiasm for our LLM Agents MOOC—12K+ registered learners & 5K+ Discord members! 📣 Excited to launch today the LLM Agents MOOC Hackathon, open to all, with $200K+ in prizes & credits! 🔗 Sign up now: rdi.berkeley.edu/llm-agents-… & join us virtually or in person @UCBerkeley! Huge thanks to our sponsors:@OpenAI @GoogleAI @AMD @LambdaAPI @Intel @SierraPlatform @OrbyAI (and more to come) 🚀 Explore 5 exciting tracks: 1️⃣ Applications: Build cutting-edge LLM agents! 2️⃣ Benchmarks: Create innovative AI agent evaluation benchmarks! 3️⃣ Fundamentals: Strengthen core agent capabilities! 4️⃣ Safety: Address critical safety challenges in AI! 5️⃣ Decentralized & Multi-Agents: Push the boundaries of multi-agent systems! Special thanks to my co-instructor @xinyun_chen_ @GoogleDeepMind & our amazing guest speakers for making this a great MOOC: @denny_zhou @GoogleDeepMind; @PercyLiang @Stanford; @8enmann @AnthropicAI; @ShunyuYao12 @OpenAI; @chi_wang_ @GoogleDeepMind; @jerryjliu0 @llama_index; @lateinteraction @Databricks; @gneubig @CarnegieMellon; @NicolasChapados @ServiceNow; @tydsh @AIatMeta; @drjimfan @NVIDIA; Burak Gokturk @Google Join us to shape the future of LLM Agents! llmagents-learning.org 🤖✨ #AI #Hackathon #LLMAgents #UCberkeley
17
106
22
386
78,767
#PropertyGPT is a very exciting work to combine formal verification with LLM. Given that we have implemented a full symbolic execution engine for solidity, we can do automatic formal verification for smart contracts with zero human intervention. Also we are working on more FM+AI.
Thanks for the shoutout @Certora! We're excited that our #AI engine #PropertyGPT is contributing to secure smart contract development 💻
1
3
613
Thanks for the shoutout @Certora! We're excited that our #AI engine #PropertyGPT is contributing to secure smart contract development 💻
Its happening @VitalikButerin PropertyGPT claims $8,256 in @code4rena contests arxiv.org/pdf/2405.02580
1
4
1
2
2,706
The 2022 IEEE Transactions on Software Engineering Best Paper Award goes to "Machine Learning Testing: Survey, Landscapes and Horizons" 🎇🎇🎇 ieeexplore.ieee.org/document…
1
6
45
2,582
This paper is finally accepted by NDSS, which is about attacking an LLM using an LLM. Seems now we are systematically learning how to "Scam" AIGC, which is both exciting and scary. #SecureAI #trustedAI #adversarialAI
3
1
653
👏👏👏 Kudos to the TOP 3 teams of 2023 MetaTrust Web3 Security CTF Competition! You have impressed us with your works. Hope we can keep exploring and BUIDL together in the Web3 security industry. #MetaTrust #CTF #Web3 #Security
5
3
6
14
10,311
As the first large-scale exposure event in 2023, @MetatrustLabs will celebrate the New Year with Web3 industry leaders. The founder Professor Liu Yang will share the topic of blockchain security from 8-12pm tonight in Twitter Space. xitter.cf/huobiglobalzh/status/1… 1/2 #metatrust
🔥Huobi闹新春,Web3之夜🔥 1月18日晚间八点,#Huobi 将携手華语区26位大咖对Web3行业进行年终盘点,马上预约直播,不要錯過:xitter.cf/i/spaces/1ynJOadgQnnKR
1
1
2
1,013
Metatrust's fully automated security audit scanning product is currently also inviting users to use it for free Registration URL: reurl.cc/85koX4 Official website: metatrust.io/ Find more information: [email protected] 2/2 #smartcontract #blockchain #Security
1
1
1
480