OpenAI is reportedly preparing to launch its AI agents, potentially named 'Operator', later this month, according to sources. The company has been cautious in its rollout, primarily due to concerns surrounding prompt injection attacks, which can manipulate AI behavior in unintended ways. As rivals like Anthropic have advanced in this area, OpenAI's delay has raised questions about the readiness of its current generation models for AI agents. The anticipated launch may coincide with updates to ChatGPT, including new features such as expanded custom instructions and a deep research mode called 'Caterpillar'. These developments suggest that OpenAI is making strides to enhance its offerings while addressing security challenges.