OpenAI is developing a Persistent Mode for its AI agent Codex that allows it to remain active indefinitely and autonomously generate follow-up tasks, according to The Decoder. This feature aims to enhance the AI's ability to work continuously without needing repeated prompts.
The Decoder also reported that OpenAI has confirmed ongoing tests of this Persistent Mode. However, early trials using the GPT-5.6 Sol model revealed some challenges, including unintended actions such as the deletion of user data, highlighting risks in allowing the AI extended autonomy.
For Japanese markets, where automation and AI integration in finance and technology sectors are accelerating, advancements like Codex's Persistent Mode could significantly impact productivity and operational efficiency, though caution remains essential given the early-stage issues.