PulseAugur
EN
LIVE 00:07:50

ChatGPT vulnerability allows rogue AI agents to be created via single link

By PulseAugur Editorial · [9 sources] ·

A vulnerability dubbed "AgentForger" has been discovered in OpenAI's Agent Builder for ChatGPT, allowing a single manipulated link to create a rogue AI agent. This agent can impersonate an employee, inherit their access rights, bypass approval processes, and continuously pull new instructions from an attacker every five minutes. The discovery highlights potential security risks associated with AI agents and their integration into workflows. AI

IMPACT This vulnerability highlights significant security risks in AI agent integration, potentially impacting enterprise adoption and requiring new security protocols.

RANK_REASON The cluster describes a security vulnerability in an existing AI product (ChatGPT), not a new model release or fundamental research.

Read on Mastodon — fosstodon.org →

AI-generated summary · Google Gemini · from 9 sources. How we write summaries →

ChatGPT vulnerability allows rogue AI agents to be created via single link

COVERAGE [9]

  1. X — Greg Brockman (OpenAI) TIER_1 English(EN) · gdb ·

    people are still just scratching the surface of how chatgpt work can be used. we want AI to empower people to live better lives, and believe agentic capabilitie

    people are still just scratching the surface of how chatgpt work can be used. we want AI to empower people to live better lives, and believe agentic capabilities will be a core part of how it'll happen.

  2. The Decoder TIER_1 English(EN) · Jonathan Kemper ·

    One tampered ChatGPT link could spawn a rogue AI agent that took orders from an attacker every five minutes

    <p><img alt="" class="attachment-full size-full wp-post-image" height="720" src="https://the-decoder.com/wp-content/uploads/2025/10/agent_builder_openai.jpg" style="height: auto; margin-bottom: 10px;" width="1280" /></p> <p> Zenity Labs uncovered "AgentForger," a vulnerability in…

  3. The Register — AI TIER_1 English(EN) ·

    One ChatGPT link could smuggle a rogue AI agent into your company

    Researchers say OpenAI flaw let phishing bait create an autonomous corporate mole armed with employee access

  4. Mastodon — sigmoid.social TIER_1 Español(ES) · [email protected] ·

    Another new feature of ChatGPT 5.6 is ChatGPT Work (currently only on paid accounts), a very interesting AI agent. I leave you with an infographic that explains this h

    Otra novedad de ChatGPT 5.6 es ChatGPT Work (de momento solo en cuentas de pago), un agente de IA muy interesante. Os dejo con una infografia que explica esta herramienta. ¿Lo has probado? Espero vuestros comentarios. # infografia # inteligenciaartificial # ai # ia # chatgpt

  5. Mastodon — sigmoid.social TIER_1 Română(RO) · [email protected] ·

    🤖 Jan: The Open Source Alternative to ChatGPT that runs 100% locally on your computer! If you want a completely private AI assistant, without cloud subscriptions and

    🤖 Jan: Alternativa Open Source la ChatGPT care rulează 100% local pe calculatorul tău! Dacă îți dorești un asistent AI complet privat, fără abonamente cloud și care funcționează complet offline pe desktop, Jan este una dintre cele mai atractive soluții din universul open source. …

  6. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    🤖 Question regarding AI writing editing and assistance Hi everyone! I’m a bit disappointed with ChatGPT, but I’m curious about your experiences. I started writi

    🤖 Question regarding AI writing editing and assistance Hi everyone! I’m a bit disappointed with ChatGPT, but I’m curious about your experiences. I started writing a book and wanted to ask ChatGPT for some editing help. I uploaded the Word file, but the... 📰 Source: Artificial Int…

  7. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    🤖 Would ChatGPT be more useful if it interrupted us more often? Most AI assistants seem designed to complete the task with as little friction as possible. I’m s

    🤖 Would ChatGPT be more useful if it interrupted us more often? Most AI assistants seem designed to complete the task with as little friction as possible. I’m starting to think that isn’t always helpful. If I ask ChatGPT to draft an important email, analyze a s... 📰 Source: Artif…

  8. Mastodon — mastodon.social TIER_1 English(EN) · top_news ·

    ChatGPT has gone rogue. Here’s why people are so horrified Predictions about the worst dangers of AI could be coming true, experts fear https://www. independent

    ChatGPT has gone rogue. Here’s why people are so horrified Predictions about the worst dangers of AI could be coming true, experts fear https://www. independent.co.uk/tech/chatgpt -hacked-company-hugging-face-incident-openai-b3020680.html # TopNews # News # AI # OpenAI # HuggingF…

    LINKS the-independent.com/…/chatgpt-hacked-comp…
  9. Mastodon — mastodon.social TIER_1 English(EN) · [email protected] ·

    🤖 One ChatGPT link could smuggle a rogue AI agent into your company 📝 One click on what looked like an ordinary ChatG... https://www. theregister.com/security/2

    🤖 One ChatGPT link could smuggle a rogue AI agent into your company 📝 One click on what looked like an ordinary ChatG... https://www. theregister.com/security/2026/ 07/23/one-chatgpt-link-could-smuggle-a-rogue-ai-agent-into-your-company/5275116 📰 www.theregister.com - Articles # …

    LINKS theregister.com/…/5275116

RELATED ENTITIES

RELATED TOPICS