By Tyler Warner
5 min read
Morning Minute is a daily newsletter written by Tyler Warner. The analysis and opinions expressed are his own and do not necessarily reflect those of Decrypt. Subscribe to the Morning Minute on Substack.
GM!
Today’s top news:
Smart contracts have lost billions to exploits.
Now the world’s most powerful AI lab and crypto’s most respected research firm are building the tools to end that.
OpenAI and Paradigm jointly released EVMbench yesterday, an open benchmarking framework that evaluates AI agents across three modes:
The benchmark draws on 120 curated high-severity vulnerabilities pulled from 40 real-world audits, mostly sourced from Code4rena competitive audit contests. It also includes scenarios from the security audit of Stripe’s Tempo, the payments-focused L1 built with input from Visa, Shopify, and OpenAI.
When Paradigm started this project, top models could exploit fewer than 20% of critical bugs. That number is now above 70%.
OpenAI also expanded the private beta of Aardvark, its dedicated security research agent, and committed $10M in API credits through its Cybersecurity Grant Program to support defensive crypto research.
Paradigm, from their EVMbench release: “It’s now clear to us that a growing portion of audits in the future will be done by agents. Hopefully this benchmark, harness, and agent serve both as a preview and an accelerant towards that future.”
OpenAI noted that measuring AI performance in “economically relevant environments is critical as models become powerful tools for both attackers and defenders.”
Smart contract exploits have drained over $5B from DeFi in the last two years alone.
EVMbench is OpenAI and Paradigm’s move to stop the drain.
The irony?
The same capability that makes EVMbench powerful (AI that can find and exploit bugs at 72% accuracy) is also the exact threat model the tool defends against. In the wrong hands, a benchmark this precise becomes a hacking playbook.
OpenAI knows this. The $10M Cybersecurity Grant and Aardvark expansion shipped alongside the benchmark for a reason.
But the more important story here is what this signals about the maturity of the AI-crypto integration thesis.
OpenAI, arguably the biggest AI lab on the planet, is formally allocating research resources to Ethereum security. They co-authored this with Paradigm and grounded it in real-world stablecoin infrastructure like Stripe’s Tempo.
This is one of the most meaningful integrations between these two worlds to date.
And don’t expect it to be the last…
Decrypt-a-cookie
This website or its third-party tools use cookies. Cookie policy By clicking the accept button, you agree to the use of cookies.