Expansion of OpenAI's API calling capabilities

OpenAI announced the expansion of their API calling capabilities, specifically introducing custom GPTs and assistance. These additions allow AI agents to call arbitrary APIs, browse the web, and even execute code. While this development has tremendous potential for humanity and business, it also poses significant security challenges. The ability of AI agents to call APIs opens up a new threat vector, potentially leading to prompt injection and the propagation of injections across the internet. With the sharing functionality of these capabilities, the possibilities for attacks become endless. Security professionals need to prepare for a world where AI agents have code execution and action-taking abilities.