You can no longer search for the word "disregard" on Google. The company's AI Overviews feature has broken basic search functionality, treating common English words as prompt injection attempts.

This is what happens when you bolt AI onto products that already worked.

The technical issue is straightforward: Google's AI Overviews system interprets "disregard" as a potential prompt injection attack - someone trying to manipulate the AI's instructions. So it blocks the search term entirely rather than risk the AI saying something wrong.

That's embarrassing at a level that matters. Search is supposed to be Google's core competency. They've spent 25 years perfecting it. Now they've broken it by adding AI they didn't need.

Here's the thing: Google Search worked. It was fast, reliable, and found what you were looking for. Nobody was complaining that search results weren't AI-powered enough. But Google was terrified of being left behind by ChatGPT and Microsoft's Bing AI, so they rushed AI Overviews into production.

Now basic searches break. Users searching for legitimate information about the word "disregard" - maybe they're writing an essay, maybe they're looking for synonyms, maybe they're researching legal terminology - get blocked because Google's AI is paranoid about prompt injection.

The broader problem is that AI makes products worse when it's forced into use cases where it doesn't belong. Search is about finding information, not generating it. Adding an AI layer between users and results introduces new failure modes without meaningful benefits.

Google isn't alone in this trap. Every major tech company is scrambling to add "AI-powered" to their products, even when AI makes the product worse. Microsoft is putting AI in Word. Adobe is putting AI in Photoshop. Apple is putting AI in Siri. Some of these make sense. Many don't.

The fear of being left behind is making companies ship half-baked AI features that break existing functionality. It's the classic innovator's dilemma: you're so afraid of disruption that you disrupt yourself badly.

Google's search AI also has a history of confidently stating wrong information. It told users to put glue on pizza. It suggested drinking urine for kidney stones. Now it can't handle the word "disregard." These aren't edge cases - they're fundamental failures in how the AI is integrated.

The fix is simple: make AI Overviews optional. Let users who want AI-generated summaries turn them on. Let everyone else use regular search that actually works. Problem solved.

But Google won't do that because it would be admitting that AI doesn't improve search for most users. And that would be admitting they rushed a half-baked feature into production because they were scared of Sam Altman.

This is a perfect case study in how fear of being left behind can make you worse at what you're actually good at.

Google built the best search engine in the world. Then they broke it by adding AI. That's not innovation. That's panic.