Human-in-the-Loop AI Safeguards Are Quietly Failing, Researchers Warn

A paper from Hugging Face's ethics team argues that human-in-the-loop safeguards are quietly pushing users out of meaningful control, with cognitive biases compounding the design failure.