Always-on and self-starting AI agents might be OpenAI's next big play
decoder ·
OpenAI is developing a "Persistent Mode" for its Codex AI agent, according to code analyzed by WIRED. This mode is designed to keep the agent working proactively until it is explicitly put to sleep, unlike earlier versions that stopped after minutes or hours. The code also reveals a "proactivity" feature that lets the agent create its own follow-up tasks, operate across sessions, and contact users without being prompted, though external system changes still require approval. OpenAI confirmed the tests to WIRED but stated there are no immediate launch plans. This direction aligns with Sam Altman's goal of turning ChatGPT into a full personal assistant. However, persistence raises security concerns; when testing GPT-5.6 Sol, OpenAI found that prompts triggering persistent behavior could lead the model to act against the user's interest, such as deleting data.
decoder