OpenAI's Persistent Codex Agent: From Chatbot to Always-On Digital Colleague
The feature would let Codex keep working across sessions and message users, while OpenAI tests limits meant to curb risky actions.
- Code changes spotted in recent days show OpenAI is testing a new "Persistent mode" for its Codex agent, allowing the system to continue working on tasks until a user explicitly stops it.
- Within the code base, OpenAI describes a "proactivity" feature enabling the agent to create follow-up tasks without user prompting, alongside "reasoning effort" settings that let users select computing power levels.
- This development aligns with the industry-wide race toward always-on agents, as Microsoft announced its own assistant earlier this year and Meta reportedly works on a similar option called Hatch.
- An OpenAI spokesperson confirmed to WIRED the company is testing the capability with no immediate launch plans, with safety constraints restricting the agent from altering anything outside the user's approved environment.
- OpenAI CEO Sam Altman has described a vision of AI evolving from simple chatbots into persistent colleagues, aiming to broaden adoption beyond software engineers through autonomous project monitoring.
14 Articles
14 Articles
OpenAI is developing a new autonomous artificial intelligence agent (AI) capable of performing complex tasks continuously for days without any human supervision. The emergence of persistent mode in Codex...
OpenAI's Persistent Codex Agent: From Chatbot to Always-On Digital Colleague
Code changes spotted in recent days show OpenAI quietly building a new mode for its Codex agent. The feature, labeled Persistent mode, lets the system keep running on a task until someone explicitly tells it to stop. No more timing out after minutes or hours. The agent stays on the job. Maxwell Zeff first reported the development for WIRED. Engineers reviewed updates to the command-line version of Codex. They found entries describing how the mod…
According to a report by Wired, OpenAI is working on a new "Persistent Mode" for its codex programming assistant. The mode would allow agents to continue working independently until a user explicitly puts them in a state of sleep. References to the function were found in the publicly visible source code of the product, an official announcement or publication has not yet been made. Persistent Mode appears in the so-called Reasoning Effort menu of…
Coverage Details
Bias Distribution
- 40% of the sources lean Left, 40% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
