TREE NEWS reports: At Devcon, ten AI coding agents attempted 12 Solidity security challenges designed for human developers. Only OpenAI’s Codex completed the course, while open-weight DeepSeek nearly finished, becoming the cheapest near-finisher. The results highlight gaps in AI’s ability to handle smart contract security tasks.
OpenAI Codex Only AI to Finish Austin Griffith’s Solidity Security Race
This result is less about raw capability and more about the economics and accessibility of AI in security-critical code. That DeepSeek, an open-weight model, nearly matched a frontier system at a fraction of cost suggests the barrier to advanced auditing is lowering, which could pressure premium AI services. It also shows that even the best agent only just cleared a bar set for humans, meaning smart contract security remains a domain where AI is a tool, not a replacement. The open question is whether these near-finishes will close the gap on more complex, real-world exploits.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.