Codex can exploit vulnerable crypto smart contracts 72% of the time, raising urgent questions about AI-powered cyber offense and defense.
OpenAI and Paradigm launched EVMbench, a tool that tests how capable AI agents are at finding and fixing smart contract vulnerabilities.
OpenAI launches EVMbench to test AI agents on smart contract security days after Claude Opus 4.6-assisted code triggered a $1.78M DeFi exploit.
Morning Overview on MSN
OpenAI unveils GPT-5.3-Codex, its first AI model trained by its own AI
OpenAI has introduced GPT-5.3-Codex, a new generation of its Codex coding system that did more than write software for others, it helped build itself. The company describes this 5.3 family model as a ...
OpenAI introduces EVMbench to measure AI crypto security. Benchmark evaluates detection, patching and exploit skills. OpenAI has launched a benchmarking system called EVMbench to evaluate how ...
OpenAI launches its OpenAI Smart Contract Benchmark called EVMbench with Paradigm to test AI agents in crypto audits.
OpenAI and Paradigm unveil EVMbench, a benchmark testing AI agents on smart contract security across 120 high-severity vulnerabilities.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results