If you write for clients, something has quietly changed this year. Agencies are reporting that close to 95% of their clients now either ask for an AI detection report or scrutinise the work for AI use before approving payment. No major freelance platform scans deliverables automatically. Clients are doing it themselves, with third party tools, after you have already done the work.
Here is the awkward part. Those tools are not reliable. One study of 715 passages found a false positive rate of almost 5% on writing that a human actually typed. That means one in twenty honest deliverables can come back flagged, and you are the one who has to argue your way out of it.
So most working writers have added a step: they scan their own work before sending it, even when they wrote every word themselves. Not because they are hiding anything, but because arguing after the invoice is due is a bad position to be in.
The bigger lesson sits underneath all of this. Around 84% of freelancers now use AI tools regularly, up from 41% three years ago. Clients are not really objecting to AI. They are objecting to writers who let AI do the thinking. The people getting rehired are the ones who use AI to move faster and still bring expertise the model cannot fake.
In the Digi Yesmads community, the freelancers who have raised rates this year are the ones who lead with the thinking, not the output. Worth adding one line to your next proposal: what you know, that a model does not.