🗞️ OpenAI teamed up with crypto firm Paradigm to launch EVMBench to test AI agents on smart contract security

2025-08-21 · Source: Rohan's Bytes · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Blockchain & Distributed Ledger Technology, Robotics & Autonomous Systems · Depth: Intermediate, medium

Summary

This edition of the daily AI newsletter, dated February 18, 2026, covers several significant developments in the AI and tech sectors. OpenAI and Paradigm introduced EVMbench, a new benchmark designed to evaluate AI agents' capabilities in smart contract security, specifically for detecting, exploiting, and patching high-severity vulnerabilities in systems holding over $100 billion in assets. Nvidia secured a multiyear deal to supply Meta with millions of AI chips, including Blackwell, Rubin, Grace, and Vera CPUs, with Grace CPUs being deployed standalone in Meta's data centers for the first time. China's AGIBOT unveiled its AGIBOT A3 humanoid robot, demonstrating advanced martial arts-style movements and mid-air maneuvers. Additionally, Figma launched a "Code to Canvas" integration with Anthropic's Claude Code, enabling the conversion of UI code into editable Figma design files rather than static screenshots. The brief also highlighted Andrew Yang's viral article, "The End of the Office," which discusses AI's potential impact on white-collar jobs and urban economies.

Key takeaway

For CTOs and AI Scientists evaluating AI's role in critical infrastructure, the EVMbench release signals a maturing capability in AI-driven security. You should investigate integrating AI agents for smart contract vulnerability detection and patching, especially given the significant performance gains seen with models like GPT-5.3-Codex. This could enhance security postures and reduce financial risks in blockchain-based systems.

Key insights

AI is rapidly advancing in specialized domains like smart contract security and UI design, while also impacting hardware markets and the future of work.

Principles

• Benchmarks drive AI agent improvement.
• AI hardware demand remains high.
• AI can automate creative tasks.

Method

EVMbench evaluates AI agents by packaging 120 high-severity smart contract bugs into detect, patch, and exploit tasks within a local Anvil EVM chain sandbox.

In practice

• Use EVMbench to test AI for smart contract security.
• Explore Figma's Code to Canvas for UI design workflows.
• Consider AI's impact on white-collar roles.

Topics

• Smart Contract Security
• AI Benchmarking
• AI Hardware
• Humanoid Robotics
• AI-powered UI Design

Best for: CTO, AI Scientist, Research Scientist, AI Engineer, AI Product Manager, Tech Journalist

Related on AIssential

• New benchmark shows AI agents can exploit most smart contract vulnerabilities on their own — AI agents demonstrate significant capability in exploiting and fixing smart contract vulnerabilities, especially with location hints.
• 😺 OpenAI built a crypto thief — AI agents, particularly OpenAI's GPT-5.3-Codex, are highly effective at exploiting smart contract vulnerabilities.
• Should you let OpenClaw pen test your system? Plus: Cybersecurity for ephemeral software — AI agents offer significant potential for cybersecurity, but require careful management and integration to be effective and safe.
• OpenAI Launches Full-Scale Effort to Patch Open-Source Bugs as It Takes on Anthropic’s Mythos — OpenAI is significantly expanding its cybersecurity initiatives and tools to address AI hacking risks.
• TAI #204: Are AI Agents Starting A Cybersecurity Arms Race? — AI agents are rapidly transforming cybersecurity, enabling both advanced attacks and unprecedented defensive capabilities.
• Our latest investment in open source security for the AI era — Industry leaders are investing $12.5 million and AI tools to proactively secure open source against AI-driven threats.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Rohan's Bytes.