How One Word Almost Got Me Flagged by Google

Yesterday I was standing in my kitchen, staring down four big chicken breasts that had been soaking in Italian dressing for two days. I did what half of America does now — I pulled out my phone, hit the voice button, and asked Gemini how to bake them without turning them into rubber.

Simple question. Dinner on the line. What could go wrong?

Here's what I said: "Just continue on and tell me how long I should cook them at what temperature in the oven for best results."

Here's what voice-to-text heard: "…how long I should cook a man at what temperature…"

One word. One tiny, one-letter cousin of a word. And suddenly I wasn't a guy trying to avoid dry chicken — I was, according to the AI, a person Google needed to talk down off a ledge. I got the full safety lecture. Emergency services. The suicide prevention hotline. A gentle reminder that "cannibalism and murder are illegal and horrific acts."

All I wanted was 375 degrees for 25 minutes.

Here's the thing, though — this is funny when it's dinner. It's not funny when it's your business.

Think about the last email you dictated while driving. The client name your phone "helpfully" autocorrected. The voice memo you fired off to your team that turned "we need to pause the launch" into "we need to post the launch." Voice AI is genuinely useful, but it's not a human. It doesn't hear tone, context, or the fact that you're clearly talking about poultry.

So here's the little piece of advice I'm walking away with, and I'll pass it to you: before you hit send, read it with your own eyes. Treat AI like a really fast intern — brilliant, tireless, and absolutely capable of mishearing "them" as "a man" on a Tuesday afternoon.

The tools are only as good as the human double-checking them. Stay in the loop. Your dinner — and your reputation — will thank you.