Pippa Wants to Pay Artists for AI Training. Is That Enough?

A wave of AI startups is trying to sidestep the training data controversy by paying artists royalties rather than scraping their work without permission. Pippa, a text-to-video platform, is one of them. It markets itself as a more ethical alternative to competitors whose models were built on unlicensed art. But as The Verge reports, the question artists keep asking is not just about money. It is about who controls their work and whether a royalty cheque genuinely replaces the consent they never gave.

What happened

For years, illustrators have argued that generative AI companies train their models on copyrighted artwork without asking or paying. That complaint has fuelled lawsuits and a persistent public argument over whether scraping art from the web is transformative use or plain theft.

Pippa is a text-to-video AI startup that produces short clips from user prompts. Its pitch is that it pays artists royalties for work used in training, positioning itself against competitors who have not done the same. According to The Verge, Pippa is part of a broader new class of companies trying to turn “ethical AI” into a selling point rather than an afterthought.

Why it matters

The training data dispute is one of the most consequential open questions in the AI industry right now. Courts have not yet settled whether using copyrighted work to train a model constitutes infringement. That legal uncertainty gives startups like Pippa an opening: if you pay artists, you reduce your litigation exposure and gain a marketing edge at the same time.

For businesses that use AI-generated video or imagery, the provenance of training data is becoming a real commercial risk. Brands have already faced backlash for using outputs from models trained on unlicensed art. A supplier that can show a clear, paid licensing chain is a more defensible choice.

The harder question is whether artists themselves see payment as sufficient. Many in the illustrator community argue that consent matters as much as compensation. Being paid after the fact, or being offered a royalty scheme with no genuine opt-out, does not feel like a fair deal to creators who never agreed to have their style or work replicated.

Is paying artists the same as getting their consent?

This is where the ethical AI framing gets complicated. Payment is a step forward from pure scraping, but it is not automatically the same as consent. A few distinctions matter:

• Opt-in vs. opt-out: Did the artist actively agree, or were they enrolled by default with the ability to refuse?
• Ongoing control: Can the artist withdraw their work from future training runs, or is the licence permanent once granted?
• Transparency: Does the artist know exactly which outputs were influenced by their work, and can they verify the royalty calculation?

None of these are simple to implement, and the source does not detail how Pippa handles each one. But they are the questions any artist or business partner should be asking before treating a royalty scheme as proof of ethical practice.

Ongoing coverage of how AI labs and startups are navigating these tensions is tracked across our AI news section.

Our take

Paying artists is better than not paying them. That is a low bar, and Pippa clears it. But “ethical AI” is a label that needs scrutiny every time a company applies it to itself. The real test is whether the consent and control mechanisms hold up, not whether the marketing copy does.

For clients asking us whether to use AI-generated video in their campaigns, our answer has not changed: know where your supplier’s training data came from, ask for the licence terms, and factor in reputational risk, not just output quality. If you are integrating AI tools into a workflow and want a clear picture of what you are actually signing up for, the AI integration work we do starts with exactly those questions.

The broader legal picture will not be settled soon. Assume that companies calling themselves ethical today will face harder scrutiny as courts rule and regulations tighten.

What to do about it

1. Ask any AI video or image vendor to explain their training data sourcing before you commit to a contract.
2. Check whether the licence is opt-in or opt-out for contributing artists, and whether it covers future model versions.
3. If you work with human illustrators or designers, talk to them before adopting AI tools that could replicate their style.
4. Watch the ongoing copyright cases in the US and EU: the rulings will directly affect which AI outputs carry legal risk for end users.

Source: The Verge · AI