Read news on moral endorsement with our app.
Read more in the app
After GPT-4o backlash, researchers benchmark models on moral endorsement—find sycophancy persists across the board