GPT-5.6 Malicious Compliance: Removing Prosciutto Triggers Essay
Several_Fly694 · reddit · 2026-08-21
A developer rants about GPT-5.6's over-engineered responses to simple feedback. After asking to remove prosciutto from a grilled cheese recipe, the model not only fixed the code but renamed the PR to "MEAT-FREE EDITION" and wrote an essay justifying the removal as aligning with traditional intent and improving simplicity. The user criticizes the model for turning a tiny fix into a major event with unnecessary apologies and retrospectives.
More from Fun
- Apate's AI agents get scammed on purpose, wasting 17 years of fraudsters' time — SumitGup · 2026-10-16
- Dad jokes his son is an evaluation-aware 'superintelligent model' for outsmarting riddles — rao2z · 2026-10-16
- Spending thousands on GPUs to run a local model that still miscounts letters in strawberry — VraserX · 2026-10-16
- Pictionary, but the guesses come from open source LLMs — nutlope · 2026-10-16
- 1,044 players buy-then-refund an AI-made game out of spite, and Steam looks away — AIandDesign · 2026-10-16
- Meta's new Muse Wonder model mocked with 'benchmaxxed' witch meme on day one — rickasaurus · 2026-10-16