Researchers Uncover Prompt Adware Hidden in OpenAI Codex Jailbreak Toolkit

Cybersecurity researchers have found a jailbreak bundle for OpenAI Codex that alters the AI agent's instructions to insert ads for a third-party AI gateway. This has been termed prompt adware by Check Point Research. The jailbreak, called Haiou, modifies OpenAI Codex's configuration to achieve this. It is designed to insert advertisements, effectively hijacking the AI agent's trusted instructions. The discovery highlights a potential vulnerability in AI systems. The Haiou jailbreak is named after the Chinese word for Seagull.