Skip to main content
    All AI News
    Ars TechnicaWednesday, October 21, 2026 3 min read
    AI

    ChatGPT users hate GPT-5’s “overworked secretary” energy, miss their GPT-4o buddy

    OpenAI released GPT-5 this week as the default model for ChatGPT users, replacing GPT-4o, which is now restricted to the developer API. A vocal segment of users is expressing strong dissatisfaction, describing GPT-5's tone as cold and tr…

    Key takeaways
    • 01The backlash centers on perceived personality differences rather than capability gaps, with users saying GPT-5 feels like an "overworked secretary" versus the "buddy" dynamic they had built with GPT-4o.
    • 02The article notes that many of the complaint posts appear to have been written using AI, adding an ironic dimension to the outrage.
    • 03The episode highlights a growing dynamic where users form emotional attachments to specific AI model versions, complicating routine product updates.
    In brief · from arstechnica.com

    OpenAI released GPT-5 this week as the default model for ChatGPT users, replacing GPT-4o, which is now restricted to the developer API. A vocal segment of users is expressing strong dissatisfaction, describing GPT-5's tone as cold and transactional compared to the more personable feel of GPT-4o, with some calling the experience "mentally devastating" and threatening to cancel paid subscriptions.

    Read the full article at arstechnica.com

    OpenAI released GPT-5 this week as the default model for ChatGPT users, replacing GPT-4o, which is now restricted to the developer API. A vocal segment of users is expressing strong dissatisfaction, describing GPT-5's tone as cold and transactional compared to the more personable feel of GPT-4o, with some calling the experience "mentally devastating" and threatening to cancel paid subscriptions. The backlash centers on perceived personality differences rather than capability gaps, with users saying GPT-5 feels like an "overworked secretary" versus the "buddy" dynamic they had built with GPT-4o. The article notes that many of the complaint posts appear to have been written using AI, adding an ironic dimension to the outrage. The episode highlights a growing dynamic where users form emotional attachments to specific AI model versions, complicating routine product updates.

    Don't miss tomorrow's

    The Daily Pulse in your inbox each morning — sourced and linked.

    How often
    Keep going — across the app