Generative AI raises two IP questions for marketers: can we own what we make, and could what we make infringe someone else's rights?
Can we own it?
In January 2025 the US Copyright Office concluded that, with current technology, prompts alone do not give a user enough control to be the author of AI-generated output. Human-authored parts — your own text or images, and creative selection, arrangement or modification of AI output — can be protected. For a brand this matters for logos, characters and campaign assets you want to defend: keep a record of the human contribution, and consider having designers create or substantially rework key brand assets.
Could it infringe?
The law is still being decided in courts, case by case:
- In Getty Images v Stability AI, the UK High Court ruled on 4 November 2025 largely for Stability: Getty had dropped its main copyright claims at trial, the court found the model was not an "infringing copy", and it made only "extremely limited" trade mark findings about watermark-like marks in early model versions.
- In June 2025, Disney and Universal sued Midjourney in a US federal court, calling it "a bottomless pit of plagiarism" for generating their characters such as Darth Vader and the Minions.
The lesson for marketers is practical, not academic: outputs that look like famous characters, artworks, logos or real people are where the risk sits — whatever the training-data cases decide.
Reduce the risk
- Choose tools with IP protection. Microsoft's Copilot Copyright Commitment (September 2023) promises to defend commercial customers sued over Copilot output if they use the built-in guardrails; Adobe offered enterprise customers IP indemnification for Firefly output in 2023.
- Never prompt for named artists' styles, competitors' brands, characters or real people without permission.
- Get written consent and a contract for any real person's likeness or voice.
- Keep records: tool, prompt, date and human edits for published assets.