Summarized by ProofAgent · Sunday, November 22, 2026
OpenAI’s GPT‑6.2 gets new prompt guide for safety

On November 18, 2026, OpenAI released new guidance for prompting its latest model, GPT‑6.2 (“Astra”). The advice recommends being concise, letting the model make safe assumptions, and warns not to edit the “prefix” of a conversation, which could break the AI’s reasoning.
Why it matters
Well‑crafted prompts help AI do what you want—understanding these can reduce miscommunications and mistakes.
What you can do
When using AI, keep your instructions simple, clear, and let the system clarify when unsure.
Read the original at Sulat AI & Tech News
This is ProofAgent's plain-English summary of reporting by Sulat AI & Tech News. Facts and quotes belong to the original source.