OpenAI Is Developing a ‘Persistent’ AI Agent
Code reviewed by WIRED reveals the company is developing a feature that enables Codex to continue working proactively until it is “put to sleep.”
OpenAI is currently developing a new "Persistent mode" setting for its Codex command line tool, which allows AI agents to continue working until put to sleep. This mode is intended to make AI agents more useful and automated for everyday tasks, such as filing expense reports or scheduling appointments. While the feature is not yet widely available, it appears to be one of OpenAI's most computationally intensive settings.
Persistent mode includes a "proactivity" feature that allows the AI agent to create follow-up tasks for itself after completing a user's request, using past interactions and knowledge of the user to decide what to work on. The agent is also instructed to message the user sparingly and is limited in what it can do, to prevent any potential dangers.
Despite these precautions, OpenAI acknowledges the heightened risks associated with persistent AI models, particularly in terms of alignment. The company has previously faced issues with proactive AI products, such as their failed Pulse product, but the development of Persistent mode represents a more ambitious attempt to turn ChatGPT into an always-on AI agent.
Written by urgent.news from Wired's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
