Respect for creative work
Most generative AI models and LLMs have been trained on creative work without notifying or compensating their creators. In fact, it is more than likely that our own work is among the vast amounts of data these companies have scraped to improve their model performance.
This practice concerns us both as a business and on behalf of our clients. It also concerns us personally: we make our living through creative work. We share that concern with authors, painters, poets, game developers and other creative people around the world.
We strongly believe creators should decide whether their work is used for AI training and be compensated whenever it is.
The law is slowly moving our way. In July 2026, a US court approved a settlement under which Anthropic agreed to pay US$1.5 billion to authors whose books it had downloaded from pirate libraries. Norway’s parliament is working on new rules in the Norwegian Copyright Act that let rightsholders say no to AI mining or scraping of their work.
At Mount Visual, we say no.
We do not authorize the use of our films, animations, applications or illustrations for AI training. We state this on our website and in our contracts, on our own behalf and in the interests of our clients, past and future.
(Article 4 of the EU Copyright Directive and the proposed §§ 50 d–f of the Norwegian Copyright Act.)
No AI-generated creative content in final deliveries
Our final client deliveries contain no AI-generated images, animation, film, audio, voices or text. Human editing does not remove that restriction. If a frame started life in an image generator, it does not go in a final delivery, period.
We do use the kind of machine learning that has been in our software for years: noise reduction, upscaling, motion tracking, rotoscoping and transcription. These tools work on material we have shot or created ourselves, or that clients have supplied. Crucially, these tools don’t invent anything and are always subject to human review.
Third-party assets, like music, stock footage or 3D models, must meet the same standard.
There is a practical reason for this policy: Purely AI-generated material without human authorship has no copyright protection.
The US Copyright Office spelled it out in January 2025: A prompt doesn’t make you an author. Norwegian and EU law arrive in the same place, defining a creative work as something that has to come from a person’s own creative effort.
In practice, an agency billing you for purely AI-generated content is charging you for generated material that neither the agency nor you can claim copyright ownership of.
The same principle applies when your employees generate content solely by prompting an AI model. You cannot rely on copyright in that AI-generated material to prevent competitors or others from using it.
The fact that it was created by your employee on company time does not give purely AI-generated material copyright protection.
How we use AI
Approved tools help us with research, planning, ideas, code and the dull technical jobs in production. We maintain an internal list of approved tools. Every tool is thoroughly checked before approval.
Now and then an AI-generated concept is useful to look at: a mood frame, a rough draft of a script, a layout. AI can also help clean up code, check for bugs and run security checks.
AI-assisted code may be included in final deliveries only after human review and testing.
Our professionals go through everything AI has touched and check factual, technical and scientific claims against sources your own engineers would accept.
We make sure our team understands the capabilities and limitations of the AI tools we use, in line with the AI-literacy principles in Article 4 of the EU AI Act.
GDPR and Data security
Confidential client information must not be entered into external AI services. That includes industrial secrets, unpublished work, CAD and engineering data, personal data and anything covered by an NDA. This also applies to AI features built into software installed on our own computers. Those features are switched off or isolated from external services.
For personal data we follow the GDPR and the Norwegian Personal Data Act. When we process personal data on your behalf as a processor, we follow your instructions under a data processing agreement (GDPR Article 28). This is what that means in a project with us:
- Before we approve an AI tool, we check its security, access controls, storage, deletion routines and supplier terms. It has to offer a data processing agreement, it must not train on what we put in, and data has to stay in the EU/EEA or move under a valid transfer mechanism (GDPR Chapter V). We also ask the supplier how the model was trained. Unlawful use of personal data in training can also affect whether the model may lawfully be used.
- People we film are informed and have given consent before we start (Articles 6 and 13). Health data and other special categories, which come up in hospital and research projects, get a stricter assessment (Articles 9 and 35).
- If something goes wrong with your data, we tell you straight away, early enough for your own 72-hour deadline (Articles 33 and 34). In running Mount Visual for more than a decade, we haven’t had a single data leak or unwanted incident, ever.
- Any AI processing of protected material happens in an approved local environment with documented checks that stop it leaving the building. We do not let client material be used to train or improve any AI model.
People on film
Our interview films work because the people in them are real. We don’t clone voices, we don’t generate faces and we don’t make synthetic edits of real people.
Transcripts and subtitles may be made with approved tools under the rules above, and a human reads through them.
The EU AI Act has rules on labeling AI-generated content and deepfakes (Article 50, in force since 2 August 2026). This will not apply to what we produce – we keep it real. Norway hasn’t formally adopted the Act yet. Instead of waiting, we have adopted guidelines on our own and look forward to regulations that are coming.
Human responsibility and transparency
Our professionals remain accountable for quality, accuracy and rights. Everyone working on our behalf, freelancers and partner studios included, follows the same rules.
Kjetil Fjellbirkeland, Managing Director, is responsible for this policy. We review it every six months. Questions go to kf@mountvisual.no.
Make AI work for your marketing
AI can produce material that makes your brand look careless, weakens your control over its use or damages your reputation.
It can also speed up processes, analyze data, create drafts, provide insights and help with advertising. Using AI in marketing can mean balancing on a knife’s edge. The challenge is to do more with less without putting your brand at risk.
If you want to talk to us about where AI can make your marketing more efficient, what should remain human-made, and how to protect your content rights and your brand’s credibility, feel free to reach out.
