Google's stated policy targets QUALITY, not authorship: helpful content ranks regardless of what drafted it; mass-produced generic filler gets demoted regardless of authorship too. The penalty risk isn't 'AI detected'. It's 'nothing added,' which AI-stuffed sites achieve at scale.
The workflow that stays safe and useful: AI drafts, you inject what only you have, real jobs, real prices, your market's quirks, your opinions, then edit into your voice and fact-check every claim. Publishing raw output is how businesses end up asserting things they've never done.
The voice trick that actually works: feed the AI your voice first. Three of your best emails or pages plus 'write like this' beats every 'sound human' instruction. And keep your no-list (words you'd never say); a brand is half vocabulary.