Larian Studios Abandons Generative AI for Divinity Development
Larian Studios confirmed it will not use generative AI on the next Divinity game-no art, no writing, not even early concept passes. The studio shifted course after community pushback made one thing clear: fans want human-made worlds and voice.
For creatives, this is more than PR. It's a public commitment to authorship, provenance, and clarity on where the work comes from.
What changed
Initial comments from CEO and creative director Swen Vincke acknowledged internal experiments with generative tools for concept exploration and placeholders. None of it was intended for the shipped game or to replace artists.
The community still pushed back. Concerns centered on ethics of training data, erosion of authorship, and the fear that "limited testing" normalizes AI in creative roles.
The new policy (clear and simple)
- No generative AI art in any form-final or concept.
- No generative AI writing for dialogue, lore, quests, or narrative.
- Possible use of non-generative AI for technical support or optimization, if ethically sourced and not replacing creative labor.
Why this matters to creatives
Trust is currency. Declaring "no gen AI" for creative assets removes ambiguity and protects the value of human style, judgment, and taste.
It also sets a precedent. If you lead a team, expect more clients and communities to ask for provenance and contribution detail-not just the final result.
Practical steps for studios and teams
- Publish a one-page AI policy: what's allowed, what's not, who signs off.
- Audit your toolchain: confirm data sources, licenses, and where outputs show up.
- Track provenance: embed creator credits and origin metadata where possible. See the C2PA standard for content authenticity.
- Set review gates: concept, pre-production, and content lock should include an "AI-origin" check.
- Update contracts: spell out acceptable tools and disclosure requirements for vendors and freelancers.
Writing and narrative: quality over quantity
Larian's writing team found generative text didn't meet their bar or improve workflow. Volume isn't the problem-voice, context, and consequence are.
If your work relies on character depth, thematic cohesion, and long-horizon callbacks, automation won't carry the weight. Strong editors and tight pipelines will.
Art direction: provenance is part of the brief
For visual teams, a no-gen-AI stance simplifies sourcing: every asset is credited to a human, full stop. That clarity helps with brand trust and long-term licensing.
If you do explore AI-adjacent tools, keep them on the non-generative side-upscalers, optimization, or assistive utilities that don't create new expressive content.
Industry signal
Larian joins a growing list of developers rethinking how generative tools fit into production. The message: protect the craft, be explicit with policy, and remove gray areas where possible.
Expect more teams to adopt provenance frameworks and public commitments around creative credit, especially on narrative-heavy projects.
What's next
No release window yet. What you can expect: Divinity's art and story will be made by people, not algorithms.
If you're exploring ethical, non-creative uses of AI inside your pipeline-automation, organization, or QA-you can scan curated options here: AI for automation.
Bottom line for creatives
- Set your policy before production starts.
- Keep provenance and credit visible.
- Use AI where it helps operations, not where it dilutes authorship.
Follow us for more video game updates. We'll keep an eye on the best pre-order deals for Divinity with our comparator.
Your membership also unlocks:
