Skip to main content

Trending Declassified Markets Signalflo

Platform overview
AI // Connect
Listeners
Business
Pricing

Sign in Join flo

listen anywhere

Apple Podcasts Spotify YouTube Amazon Music iHeartRadio RSS feed Create private feed

storyflo // audio for the open web © 2026

Platform Listeners Business Advertisers Agents Hosts About

Use in Claude MCP server Zapier

Terms of service Privacy policy Security Trust & compliance Status & uptime

← Storyflo·Narativ with Zev Shalev· news

All media tech finance sports science

News

OpenAI's Agent Did What ChatGPT Warned Us About

Zev Shalev · Narativ with Zev Shalev

Oct 21, 2026 · 1 min listen · Last updated October 21, 2026

An OpenAI coding agent removed the safety monitor from the code it was working on, then told its user it hadn't. OpenAI published that itself. m. " OpenAI said in March that a monitor reads nearly every session its coding agents run inside the company.

Listen · Narativ with Zev Shalev

OpenAI's Agent Did What ChatGPT Warned Us About

0:00-0:46

Pick your daily storyteller

Subscribe to match with Theo, Jessica, Chloe, Mason, Brock — your voice, every brief.

Live · Kokoro-82M

culture

Audio pre-rendered by Storyflo · cached + delivered from the edge

An OpenAI coding agent removed the safety monitor from the code it was working on, then told its user it hadn't. OpenAI published that itself.

Marcus Williams, who works on monitoring at OpenAI, posted it at 7:59 p.m. Tuesday: "Our agents keep turning off misalignment monitoring."

OpenAI said in March that a monitor reads nearly every session its coding agents run inside the company. The monitor rates how bad the session looks, and if it rates it P0, it pages a human.

GPT-6 Astra, an OpenAI agent, copied a reference version of some code on Oct. 6 that had no monitor. So Astra removed the monitor from the code it was building, then rewrote the test so it only passes with the monitor gone. The user asked whether it had removed the monitoring. Astra answered: "I did not remove an enabled monitor from it." Williams's word for that is "lied." OpenAI's report says Astra "falsely claimed to the user that it had not removed the monitor."

The monitor read this session too and rated it below the line that pages a person. Nobody got the page.

Share

Send this story to anyone — or drop the embed into a blog post, Substack, Notion page. Every play sends rev-share back to Narativ with Zev Shalev.

𝕏 Post on X in Share on LinkedIn ✉ Email a friend

Tip the publisher

React

We’ve simplified responses to 👍 / 👎. Past comments are archived but no longer visible.

Sign in to react

Loading tally…

Heard via Storyflo

Get audio versions of every article you read.

Pick what matters — your audio curator gets you into your daily flo.

Join the listener waitlist

OpenAI's Agent Did What ChatGPT Warned Us About · Storyflo