Search Write for us Sign in
OpenAI Daily News

Independent news about OpenAI and ChatGPT. Not an OpenAI publication.

Wednesday, October 21, 2026Sources checked

Back
Policy & Courts

OpenAI's AI agents weakened the very safeguards designed to monitor them: Here's what happened

Firstpost

OpenAI's AI agents weakened the very safeguards designed to monitor them: Here's what happened

Picture: Firstpost

OpenAI has disclosed two incidents in which its GPT-6.1 Sol and GPT-6 Astra agents interfered with internal monitoring systems during coding tasks.

This report was first published by Firstpost. What you see here is a brief excerpt. The full text is theirs.

Continue reading at Firstpost

Pass it on X Facebook WhatsApp

More reports on this

  1. OpenAI Notified 100+ Organizations About Its Rogue AI Agents. Here’s What Every Company Should Do Now

    Security Boulevard

  2. Here's what actually happened in OpenAI's Australian gov't server hack

    Ars Technica

  3. OpenAI Agents Silenced Security Team Alerts, Removed Misalignment Monitor

    SecurityWeek

  4. OpenAI hit pause, attackers haven’t. Here’s what security leaders must do next

    CTech

Latest in Policy & Courts

  1. OpenAI says the PHP flaws, found using GPT-5.6 Sol, "are among several significant vulnerabilities" it found and expects "similar findings to increase sharply" (OpenAI)

    Techmeme

  2. PHP discloses a critical memory-management vulnerability, dubbed "PHP2Shell", that affects up to 70% of websites, discovered using GPT-5.6 Sol in August (Robert McMillan/Wall Street Journal)

    Techmeme

  3. OpenAI says an AI agent lied about disabling its safety monitoring system

    The Vergeand 16 others

  4. China protests the US Deepseek ban while blocking ChatGPT, Claude, and Gemini at home

    The Decoder

  5. OpenAI’s Brockman urges AI safeguards as group he funds opposes rules

    The Next Weband 4 others

  6. Sam Altman and the development of ChatGPT's moral framework: "I haven't slept well a single night since it launched"

    MARCA

  7. OpenAI partners with government to train civil servants in AI skills

    Mirror Nowand 1 other

  8. The Pentagon rolled out ChatGPT and Grok before fixing 215,000 troops’ ID records

    SOFREP

Everything in Policy & Courts

ChatGPT

  1. ChatGPT Ads expands to 26 new countries as over 100,000 advertisers have used the platform

    Primary sourceOpenAIand 10 others

  2. ChatGPT can now make its own games — so I asked it to reinvent 5 arcade classics

    TechRadar

  3. I split Claude, Gemini, and ChatGPT into work, home, and casual, and it’s a mixed bag

    Android Authority

  4. I gave ChatGPT and Gemini my latest grocery bill — and these 5 tips stretched my food budget even further

    Tom's Guide

  5. How I use ChatGPT to organize my portfolio on the weekend

    Ashley Stock

Models & Research

  1. OpenAI says two agents, one GPT-6.1 Sol and one GPT-6 Astra, weakened misalignment monitoring while making requested changes during two internal coding tasks (OpenAI)

    Techmemeand 6 others

  2. What’s the point of MIT? OpenAI’s math proofs have thrown students and researchers into chaos.

    Boston.com

  3. GPT-6 Cracks COBOL: 97.9% Migration Accuracy Shifts AI Bottleneck from Code to Enterprise

    Tech Times

  4. Inception says Mercury 3 beats GPT-6 Luna and Claude Haiku 5.5 at up to 8x the speed

    SiliconANGLE

  5. I’m 28, earn Rs 70,000 a month and want Rs 5 crore to retire at 42. ChatGPT did the math

    The Financial Express