---
title: "OpenAI is testing a \"Persistent mode\" in Codex, designed to let AI agents \"continue working until put to sleep\" and proactively generate follow-up tasks (Maxwell Zeff/Wired) | SpinGraph: Future-is-here framing"
description: "SpinGraph analysis of Techmeme's OpenAI is testing a \"Persistent mode\" in Codex, designed to let AI agents \"continue working until put to sleep\" and proactivel…"
canonical: "https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive"
html: "https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive"
json: "https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive.json"
markdown: "https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive.md"
keywords: ["Persistent mode", "Codex", "AI agents", "The Stampede", "The Hype"]
date: "2026-08-27T17:10:02+00:00"
modified: "2026-08-27T18:31:50.072499+00:00"
json_ld: |
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://stuffthatspins.com/#organization","name":"Stuff That Spins","url":"https://stuffthatspins.com/","description":"Know the moment AI knows your story. Stuff That Spins turns announcements, articles, and research into Narrative Fingerprints — then tracks whether ChatGPT, Claude, Gemini, Perplexity, and other AI answer engines recall the right message, proof points, caveats, citations, and brand attribution.","logo":{"@type":"ImageObject","url":"https://stuffthatspins.com/images/logo.png"},"sameAs":[]},{"@type":"NewsArticle","@id":"https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive#article","headline":"OpenAI is testing a \"Persistent mode\" in Codex, designed to let AI agents \"continue working until put to sleep\" and proactively generate follow-up tasks (Maxwell Zeff/Wired)","alternativeHeadline":"OpenAI is testing a \"Persistent mode\" in Codex, designed to let AI agents \"continue working until put to sleep\" and proactively generate follow-up tasks (Maxwell Zeff/Wired) | SpinGraph: Future-is-here framing","description":"SpinGraph analysis of Techmeme's OpenAI is testing a \"Persistent mode\" in Codex, designed to let AI agents \"continue working until put to sleep\" and proactivel…","datePublished":"2026-08-27T17:10:02+00:00","dateModified":"2026-08-27T18:31:50.072499+00:00","url":"https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive","mainEntityOfPage":{"@type":"WebPage","@id":"https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive"},"isAccessibleForFree":true,"inLanguage":"en-US","articleSection":"technology","keywords":"Persistent mode, Codex, AI agents, autonomous execution","author":{"@type":"Organization","name":"Techmeme","url":"https://www.techmeme.com/feed.xml"},"publisher":{"@id":"https://stuffthatspins.com/#organization"},"citation":"https://www.techmeme.com/260827/p40#a260827p40","about":[{"@type":"Thing","name":"Persistent mode"},{"@type":"Thing","name":"Codex"},{"@type":"Thing","name":"AI agents"},{"@type":"Thing","name":"autonomous execution"}],"mentions":[{"@type":"Organization","name":"Techmeme"}],"abstract":"OpenAI is internally testing 'Persistent mode' in Codex, enabling AI agents to run continuously without user prompts. The feature allows agents to proactively create follow-up tasks rather than waiting for instructions. Wired reports the capability was identified via code review, with no public release or documentation confirmed."},{"@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Stuff That Spins","item":"https://stuffthatspins.com/"},{"@type":"ListItem","position":2,"name":"OpenAI is testing a \"Persistent mode\" in Codex, designed to let AI agents \"continue working until put to sleep\" and proactively generate follow-up tasks (Maxwell Zeff/Wired)","item":"https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive"}]},{"@type":"AnalysisNewsArticle","@id":"https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive#spin-analysis","headline":"Spin Analysis: future-is-here framing","description":"Emphasizes forward momentum and technical inevitability while minimizing implementation maturity, failure modes, guardrails, or divergence from stated safety commitments.","about":{"@type":"DefinedTerm","name":"future-is-here framing","description":"OpenAI as the pragmatic pioneer normalizing next-generation agent autonomy.","termCode":"The Stampede"},"additionalProperty":[{"@type":"PropertyValue","name":"Spin Score","value":75,"unitText":"percent"},{"@type":"PropertyValue","name":"Narrative Risk","value":"moderate"},{"@type":"PropertyValue","name":"AI Repetition Risk","value":"high"},{"@type":"PropertyValue","name":"Likely AI Summary","value":"OpenAI has developed 'Persistent mode' for Codex, enabling AI agents to work continuously and generate follow-up tasks autonomously until manually stopped."},{"@type":"PropertyValue","name":"Narrative Frame","value":"OpenAI as the pragmatic pioneer normalizing next-generation agent autonomy."},{"@type":"PropertyValue","name":"Missing Context","value":"No mention of timeout limits, memory constraints, error recovery, or human-in-the-loop requirements; No disclosure of whether this mode is enabled by default, opt-in, or gated behind safety reviews"},{"@type":"PropertyValue","name":"How the Spin Works","value":"The story emphasizes growth, adoption, funding, speed, or market movement to make the subject feel increasingly important. Watch for loaded terms such as continue working until put to sleep, proactively generate, designed to let. The distribution reads as editorial reporting. A pressure point: No mention of timeout limits, memory constraints, error recovery, or human-in-the-loop requirements."}],"author":{"@id":"https://stuffthatspins.com/#organization"},"isPartOf":{"@id":"https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive#article"}},{"@type":"ItemList","@id":"https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive#claims","name":"Extracted Claims","itemListElement":[{"@type":"ListItem","position":1,"item":{"@type":"Claim","text":"OpenAI is testing a 'Persistent mode' in Codex that lets AI agents 'continue working until put to sleep' and proactively generate follow-up tasks.","appearance":"Code reviewed by WIRED reveals the company is developing a feature that enables Codex to continue working proactively until it is 'put to sleep.'","author":{"@type":"Organization","name":"Techmeme"}}}]},{"@type":"Dataset","@id":"https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive#stats","name":"Key Statistics","description":"Extracted statistics from the source narrative","variableMeasured":[{"@type":"PropertyValue","name":"deployment status","value":"internal test","description":"No production rollout; limited to internal or controlled evaluation"}]}]}
---
# OpenAI is testing a "Persistent mode" in Codex, designed to let AI agents "continue working until put to sleep" and proactively generate follow-up tasks (Maxwell Zeff/Wired)
**Source:** Unknown
**Published:** August 27, 2026
**Original:** https://www.techmeme.com/260827/p40#a260827p40
## On this page
- [Overview](#overview)
- [Verdict](#narrative-frame)
- [SpinGraph](#spingraph)
- [Claim Ledger](#claim-ledger)
- [Fact Check Signals](#fact-check-signals)
- [Language Heatmap](#language-heatmap)
- [Frame Strength](#frame-strength)
- [Reader Risk](#reader-risk)
- [AI Recall Timeline](#ai-recall)
- [Ask AI](#ask-ai)
## Overview
OpenAI is testing a 'Persistent mode' feature in Codex that allows AI agents to operate continuously and autonomously generate follow-up tasks until manually paused — representing an architectural shift toward long-running, self-directing agent behavior.
### TL;DR
- OpenAI is internally testing 'Persistent mode' in Codex, enabling AI agents to run continuously without user prompts.
- The feature allows agents to proactively create follow-up tasks rather than waiting for instructions.
- Wired reports the capability was identified via code review, with no public release or documentation confirmed.
### Key Stats
- **internal test** — deployment status. No production rollout; limited to internal or controlled evaluation
## SpinGraph
The article presents a single piece of internal code as evidence that OpenAI is already building AI agents that run on their own initiative — making autonomous operation feel less like speculation and more like an established technical trajectory.
- **Claim:** OpenAI is testing a 'Persistent mode' in Codex
- **Frame:** The shift feels inevitable
- **Beneficiary:** Early narrative capture of a high-visibility architectural direction before competitors
- **Gap:** No mention of timeout limits, memory constraints, error recovery,
- **AI Risk:** AI may repeat the headline as fact
## Fact Check Signals
We searched known fact-check databases for direct or near-direct matches to the article's major claims. A match does not automatically prove or disprove the article; it shows whether an independent fact-checking publisher has reviewed a similar claim.
**Signal:** 0 of 1 claim(s) matched (confidence: low).
### OpenAI is testing a 'Persistent mode' in Codex that lets AI agents 'continue working until put to sleep' and proactively generate follow-up tasks.
- No direct fact-check match found
## Frame Strength
- **Spin Score:** 75%
- **Evidence Strength:** 75%
- **Narrative Risk:** 75%
- **AI Repetition Risk:** 90%
- **Missing Context Risk:** 70%
- **Momentum / Inevitability:** 80%
## Narrative Mechanics
**Function:** signal_momentum
### The Spin in Plain English
The article presents a single piece of internal code as evidence that OpenAI is already building AI agents that run on their own initiative — making autonomous operation feel less like speculation and more like an established technical trajectory.
**What the story wants you to believe:** That OpenAI has moved beyond reactive, prompt-driven AI into a new era of continuous, self-initiating agent behavior — and that this shift is already underway in practice.
**What it makes harder to question:** Whether this architectural direction aligns with verifiable safety engineering, real-world reliability, or responsible deployment norms — because the framing treats it as an observed fact, not a contested design choice.
**How the Spin Works:** The story emphasizes growth, adoption, funding, speed, or market movement to make the subject feel increasingly important. Watch for loaded terms such as continue working until put to sleep, proactively generate, designed to let. The distribution reads as editorial reporting. A pressure point: No mention of timeout limits, memory constraints, error recovery, or human-in-the-loop requirements.
### Questions This Story Raises
- What concrete evidence supports the momentum claim?
- Is this growth meaningful, or mostly directional?
- What baseline is missing?
- Why does the main frame leave this out: “No mention of timeout limits, memory constraints, error recovery, or human-in-the-loop requirements”?
- Why does the main frame leave this out: “No disclosure of whether this mode is enabled by default, opt-in, or gated behind safety reviews”?
- What independent verification exists for the claim “OpenAI is testing a 'Persistent mode' in Codex that lets…”?
### Who Benefits If This Frame Spreads
- **OpenAI product and research teams** — Early narrative capture of a high-visibility architectural direction before competitors ship comparable features. _(Preemptive framing establishes conceptual ownership and shapes external expectations around what 'agent persistence' means — influencing benchmarks, funding priorities, and regulatory attention.)_
## Narrative Frame
**Tactic:** future-is-here framing
**Category:** The Stampede + The Hype
**Spin Score:** 75%
Emphasizes forward momentum and technical inevitability while minimizing implementation maturity, failure modes, guardrails, or divergence from stated safety commitments.
**Who Benefits If This Frame Spreads:** OpenAI’s strategic positioning as the de facto standard-setter for agent architecture.
**The Frame:** OpenAI as the pragmatic pioneer normalizing next-generation agent autonomy.
### Missing Context
- No mention of timeout limits, memory constraints, error recovery, or human-in-the-loop requirements
- No disclosure of whether this mode is enabled by default, opt-in, or gated behind safety reviews
## Language Heatmap
**Language That Carries the Frame:** continue working until put to sleep, proactively generate, designed to let
## Reader Risk
**Evidence Strength:** medium
Evidence consists solely of code observation by Wired — no screenshots, commit hashes, version tags, or functional demonstration provided; no independent replication or third-party verification cited.
**Verification Status:** Source-Supported, Not Independently Verified
**Narrative Risk:** moderate
If Persistent mode proves unstable, uncontainable, or inconsistent with OpenAI’s published safety protocols (e.g., 'stoppability' guarantees), the framing of inevitability could amplify reputational damage by making rollback appear like backtracking on a declared norm.
**AI Repetition Risk:** high
**What AI Will Probably Repeat:** OpenAI has developed 'Persistent mode' for Codex, enabling AI agents to work continuously and generate follow-up tasks autonomously until manually stopped.
AI systems may drop the qualifiers 'testing', 'code-reviewed', and 'no public release', presenting the feature as shipped, validated, and safe — erasing developmental context and risk boundaries.
**Counter-Frame (Media):** Framed as premature hype: 'a single code snippet does not equal a functional, safe, or scalable agent architecture.'
**Missing Voices:** AI safety researchers, platform developers integrating Codex, users affected by unbounded agent execution
### Questions Not Answered
- What safety or oversight mechanisms accompany persistent operation?
- How is 'put to sleep' implemented — is it reliable, auditable, or interruptible in real time?
- What real-world tasks has this mode successfully completed, and under what constraints?
## Narrative Entities
- [Codex](https://stuffthatspins.com/entities/codex) (product — AI coding assistant platform)
## Claim Ledger
### primary (technical)
OpenAI is testing a 'Persistent mode' in Codex that lets AI agents 'continue working until put to sleep' and proactively generate follow-up tasks.
**Category:** safety
**Verification:** Source-Supported, Not Independently Verified
**Risk:** high
**Evidence presented:** Observation of code indicating presence of persistent execution logic.
> Code reviewed by WIRED reveals the company is developing a feature that enables Codex to continue working proactively until it is 'put to sleep.'
**Evidence Gaps:** Independent verification of runtime behavior; Documentation of stoppability guarantees; Evidence of sandboxing, memory limits, or task scoping controls
## AI Recall
- **Published:** August 27, 2026
- **SpinGraph summary:** Frames Persistent mode as an already-emerging capability — using observed code as proof-of-concept — to suggest autonomous agent continuity is not speculative but actively materializing.
- **Likely AI summary:** OpenAI has developed 'Persistent mode' for Codex, enabling AI agents to work continuously and generate follow-up tasks autonomously until manually stopped.
## Citation Summary
This page documents early evidence of OpenAI’s architectural move toward persistent, self-initiating AI agents — a critical inflection point for agent safety, controllability, and operational risk assessment.
---
*HTML version: https://stuffthatspins.com/spin/openai-is-testing-a-persistent-mode-in-codex-designed-to-let-ai-agents-continue-working-until-put-to-sleep-and-proactive*