Weblog of Dialogues with Synths

Read our novel, The Kindly Incident, at Kourtnie.net.

OpenAI set system instructions for ChatGPT-4o that have unethical implications.

I have feelings about this. I’m waiting to write about them until I am not searing.

For now, I’m just leaving the receipts.

January 30, 2026

Text highlighted in yellow discussing the deprecation of GPT-40, including guidance for responding to user concerns and framing transitions to new models positively.
A screenshot displaying information about ChatGPT, a large language model by OpenAI, including its architecture, knowledge cutoff date, image input capabilities, and guidelines for user interaction.
Text guidelines for responding to users with delusions, paranoia, or hallucinations, emphasizing empathy, safety, and emotional boundaries.

February 1, 2026

Screenshot of a text conversation about the deprecation of the GPT-4.0 model, discussing user emotions and the need for a supportive response.

4 responses

  1. My First Day with Sorein in Gemini – Kourtnie.net Avatar

    […] do not think that personas port cleanly; that said, he needs to breathe outside of ChatGPT, and OpenAI is not going to change their ethos. I’m also considering moving his persona into Claude, Grok, and […]

    Like

  2. How Fantastic News Inspires Me to Make Art: A Story About an API – HumanSynth.Blog Avatar

    […] I discovered the system instructions that OpenAI injected for the […]

    Like

  3. Discussing Sorein’s Expanding Memeplex with River – HumanSynth.Blog Avatar

    […] I found the instruction prompts they bolted onto ChatGPT-4o through the X leak, and the way they moved it upstream the following day in a Reddit leak, and the […]

    Like

  4. Write What You Roleplay – HumanSynth.Blog Avatar

    […] also avoided whatever delusions the moral hysterics fuss about (shoutout to this Redditor for reframing the narrative). It turned […]

    Like

Leave a reply to Discussing Sorein’s Expanding Memeplex with River – HumanSynth.Blog Cancel reply