Scroll Facebook now and the generated pictures announce themselves before the caption loads: high even light, skin polished to a soft plastic, the same shallow blur behind a quilted handbag or a birthday cake. That sameness is the signature of Emu, Meta’s image model, and Meta has wired it into the feed itself. Brand imagery now competes with the platform’s own generator and gets redrawn in its taste, flattening the distinct look a house pays years and a media budget to own.

The generator did not stay in a separate app. At Connect 2024, Meta pushed its Imagine tool into the feed, letting anyone type a prompt to make pictures inside their feed, Stories, or profile. It also began testing a row of AI-generated images tuned to a user’s interests, nudging them to reimagine the photo already in front of them. The model generates high-resolution images from text prompts, which is what gives its output that even, weightless finish. A shopper can regenerate your product shot with a swipe, and the version they make will read as Meta, not as you.

On the paid side the redraw is already routine. Meta’s Advantage+ creative invents new backgrounds around a product and stretches images to fill every placement, and the case studies sell the lift and skip the loss: beauty brand Fresh booked a 5x incremental return on ad spend and 42% incremental conversions once the background stopped being its own. The stated destination goes further still. As the Wall Street Journal first reported and eWeek recounted, by late 2026 a business should be able to hand Meta a product image and a budget and get the whole ad back: imagery, video, copy, audience, and a cut tuned to each viewer. Point that machine at a coat, and the art direction leaves your studio for Meta’s model.

None of this is a pilot. Meta counted 4 million advertisers on its generative creative tools at the end of 2024, up from one million six months earlier, and told investors the figure had reached 8 million by the first quarter of 2026, most of them small and medium-sized businesses. When eight million advertisers reach for one generator, their pictures converge, and the point of convergence is Emu’s taste, not theirs.

Fashion has run this experiment before, on cloth. Through the 1950s, Paris houses licensed their toiles, the muslin blueprints of a design, to American stores that cut legal line-for-line copies. A Dior silhouette an atelier needed a season to resolve could reach a sales floor as a pattern anyone could stitch. A copyist could match the corolle line of the New Look and still miss the canvas interlining and the weight of wool that held the shape away from the body. The buyer who cared felt that difference in the seams.

Online, there are no seams to feel.

In the feed the image is the whole product, and a generated one carries no interlining to give it away. When the platform hands you the silhouette, the light, and the drape, brand distinction shrinks to what a prompt cannot fetch — a color you mixed and hold, a proportion you return to until people read it as yours, a point of view the model has no reference for. Everything else Emu can approximate by Tuesday. Brands spent years teaching audiences to know them on sight, and the feed is now teaching those same audiences that everything looks like this.