Netflix admits it: around 300 of its titles this year went through generative AI

Netflix released its quarterly shareholder letter on 16 July. Between the revenue and the margins sits one paragraph on generative AI. It says more than the tables around it - because for the first time the company puts a number on something it had long talked around.

Behind the shine of 'most-watched premiere' there's a second scene, and it isn't facing the viewer. This paragraph is written for shareholders. The language gives it away - 'faster and cheaper'. This isn't a story about creativity, it's about margin. The number 300 is dropped in passing to signal to Wall Street that costs are under control, while on top it's dressed up as 'expanding what creators can do'. I've written announcements like this - I know when a number is for the press and when it's for the investor. This one is for the investor. The actual use, honestly, is boring and legitimate: touching up a crowd in post is a technique, not a revolution. What's interesting is that they chose to brag about it now, and right where the margin gets counted.

The number missing from the letter is the other important one: of those roughly 300 titles, how many told the viewer directly that AI was used. On screen - as far as it shows - none of them. That's where the weight falls for smaller players like us. When you put generative AI into client work, spell out where and how right from the start. The transparency a big player can skip toward the public and address only to the stock market is, for an agency at our scale, the exact opposite - it's the currency of trust. Netflix can stay quiet and people will still watch. We can't.

The visual is generated code-art, no third-party images.