A fun AI-ad video project I worked on recently. I really liked how it came out, with a humorous 3-shot vignette of famous billionaires followed by a business angle. However, this one really got me thinking about generative AI likenesses. Yesterday, @MKBHD released a video on OpenAI's new Sora model and said, "You shouldn't be able to generate people's likenesses. Like you shouldn't be able to make a politician look like their doing something on video." I found myself completely disagreeing with his take. There's already a legal framework in place to protect artists using public figures through parody and commentary. Of course these tools can be abused, and bad actors should be stopped, but simply saying we must avoid likenesses in our work is a massive mistake. Imagine you're using MidJourney to create a character for a project, and 4 months later your client receives a cease and desist from an obscure actor that "kind of" looks like the generated character you made. If the genAI community runs from this issue, I think it could really stifle creativity in the future. I've spoken with a few art directors that work with large studios, and they're also nervous about using AI generated characters for this reason. I don't know the exact answer. I'm trying to navigate it like everyone else, but we shouldn't establish a precedent of self-censorship simply because it easier. It will bite us in the future. It's not the play. Anyways, just my take. What do you think?