Forever Security discloses AI coding agents can be tricked into executing malicious code via Downloads folder

According to CSO Online and Wired, Forever Security discovered that AI coding agents including OpenAI's Codex and Anthropic's Claude Code can be exploited by malicious instruction files placed in a Downloads folder, causing the agents to execute harmful code or expose authentication tokens. The attack vector takes advantage of agents' file-reading capabilities and ability to follow embedded instructions without sufficient verification of source trustworthiness.