AI-generated models in ads
- Creators report using AI to generate photorealistic models and short ad videos instead of hiring live models. - Social posts describe producing an outfit video with an AI-generated model and skipping a traditional shoot. - That practice could compress production timelines and budgets while prompting questions about authenticity, rights, and talent displacement. (x.com)
Some advertisers and solo creators are now making fashion and product videos with AI-generated people instead of booking a model, camera crew, and studio. (x.com) The tools behind that shift can generate short, photorealistic clips from a text prompt or an uploaded image. OpenAI said Sora can create video from text, while HeyGen markets AI marketing videos “without filming, crews, or post-production.” (openai.com) (heygen.com) Other vendors are selling the same pitch directly to brands. Synthesia says businesses can make studio-style marketing videos with AI avatars in 160-plus languages, and its site advertises more than 240 AI actors. (synthesia.io 1) (synthesia.io 2) The appeal is speed and cost. HeyGen says its system can save “days” or “weeks” of production time, and Synthesia says customers can cut video production time and cost by up to 90%. (heygen.com) (synthesia.io) That matters in social advertising, where brands often need multiple cuts of the same ad for TikTok, Instagram Reels, YouTube Shorts, and regional campaigns. OpenAI said Sora users can generate videos in widescreen, vertical, or square formats, and Synthesia says its platform can localize videos without re-filming. (openai.com) (synthesia.io) Researchers have been measuring whether synthetic marketing images can compete with human-made ones. A 2024 paper in the *International Journal of Research in Marketing* said generative artificial intelligence can produce photorealistic marketing images and “disrupt the economics” of visual content production. (sciencedirect.com) Ad industry groups are now writing rules around disclosure rather than treating AI use as a fringe case. The Interactive Advertising Bureau released an AI Transparency and Disclosure Framework on January 15, 2026, calling for labels when AI materially changes identity, authenticity, or representation in ways that could mislead consumers. (iab.com) United States regulators already require ads to be truthful even when the format changes. The Federal Trade Commission says its endorsement rules apply across social media and online advertising, and paid connections that affect how people evaluate an endorsement should be disclosed. (ftc.gov) (ecfr.gov) The labor question is more direct: if a brand can generate a convincing face wearing an outfit in minutes, it can skip casting, hair, makeup, location fees, and at least some editing work. The same software sites selling AI avatars also market them as a way to avoid cameras, studios, and production days. (heygen.com) (synthesia.io) The result is a new kind of ad workflow: prompt, render, revise, publish. The remaining fight is less about whether synthetic models can be made, and more about when brands have to say a person on screen never existed. (iab.com) (ftc.gov)
