3 min read
News Brief Depi's Research Team reported a chain against openai/codex that began with a prompt injection in a GitHub issue and ended in a persistent backdoored workflow. Bugcrowd lists it as accepted and fixed, with a Bugcrowd priority of P1.