# Creative AI Ethics

**Track:** Model Foundations — AI for Entrepreneurship — model foundations (5)
**Framework / surface:** venture strategy
**Level:** Intermediate
**Prerequisites:** Open vs. Closed Models, Ownership, Licensing & Provenance

**In one line:** Training-data consent, style mimicry, and machine authorship — the ethics that arrive as lawsuits, licenses, and credits, and the studio stance that survives them.

## Theory, aesthetics & inspiration

The ethics of creative AI arrive concretely — as lawsuits, licenses, and credits. Training data is the first front: Getty Images sued Stability AI in 2023 over scraped photographs — its U.K. claims largely failed at trial in 2025, while its U.S. case continues — and the artists' class action Andersen v. Stability is still testing whether ingestion for training infringes. Style is the second: style as such is not copyrightable, yet prompting "in the style of" a living, named artist can drain the very market that artist built. Authorship is the third: the U.S. Copyright Office's 2023 guidance holds that material generated wholly by a machine is not registrable — human authorship must be shown, a line drawn in practice in the Zarya of the Dawn decision. Provenance standards like C2PA content credentials let work carry its own history. The durable stance for a studio is consent and disclosure as defaults — not because the law is settled, but because it is not.

**Founder question:** If the artist whose style you're borrowing were in the room, would your workflow survive the conversation?