Google removes AI Overviews for certain medical queries

This isn't a bug; it's the feature working exactly as designed. Google shipped a hallucination engine and hoped the brand would carry it through. Turns out doctors, recipes, and medical queries don't forgive confidence plus wrongness.

The Guardian didn't discover a glitch—they found the floor of what LLM-over-retrieval actually does at scale. Google has zero guardrails between "it sounds plausible" and "publish it to 100 million people."

For client work, this is the reminder that slapping an LLM onto a customer-facing surface requires domain expertise and testing infrastructure, not just API calls.