OpenAl is sticking to its story that it never intended to copy Scarlett Johansson’s voice when seeking an actor for ChatGPT’s “Sky” voice mode.
This all “feels personal," the voice actress said, "being that it’s just my natural voice and I’ve never been compared to her by the people who do know me closely.”
This comes at a time when many studios are otherwise intrigued by the idea of using AI for things like digital effects but remain, after a long history of avoiding copyright conflicts, hesitant to connect with any company potentially viewed as stealing artists’ work without consent, Reuters reported.
Setting aside whether soundalikes-hyped-as-the-real-deal is a violation of personal likeness rights…
How do we know what voice(s) they actually used? To my understanding, the process atomizes the input such that you can never actually prove what went into it.
Their whole business seems to be one of selling plausible-deniability engines.
I think a huge issue currently is the widespread “AI” hysteria. People kinda want to believe that they did this violation because they already have this negative image of LLMs in their mind from all the overblown headlines & scenarios that they’ve read.
I think it’s fair to err on the side of shadiness with OpenAI considering their training data controversies with gtp and Sora.
I don’t mean this argumentatively but would you be surprised if it came out that they did use Johansson’s voice to train it?