Security company BlockSec retested EVMBench and concluded that the benchmark overestimated AI's automation capabilities in smart contract auditing. They expanded the testing to 26 model configurations and introduced 22 real attack incidents that occurred after February 2026. The results showed that in 110 test cases, AI had a 0% success rate in exploiting real attacks, but its performance in vulnerability detection was similar to the original report, with some models able to identify known pattern vulnerabilities. (The Block)