Veracode reports Claude Opus 5.5 and GPT-6-Sol security gains

According to a new report by Veracode, overall AI-generated code security improved for the first time in a year, with vulnerability rates dropping from 44% to 38% and overall security jumping from 56% to 62%. Anthropic's Claude Opus 5.5 and OpenAI's GPT-6-Sol led the rankings with a 73% pass rate, though researchers noted that the gains were primarily driven by easy fixes for cross-site scripting while complex data-flow reasoning flaws remained challenging. Open article →