Scenes — Code of Conduct for Responsible AI
Copenhagen, last updated 9 September 2026
At Scenes, artificial intelligence is a production tool within a human-led creative process. Our work combines professional art direction, photography, AI-generated elements, editing and post-production.
We use AI to expand what brands can create without compromising quality, transparency, intellectual property, privacy or public trust.
Compliance and responsibility
Scenes is committed to complying with applicable Danish and European legislation, including the EU AI Act, the Danish Marketing Practices Act, data protection law, copyright law, trademark law and relevant contractual obligations.
Scenes generally acts as a user—or “deployer”—of third-party AI systems rather than as a developer of the underlying AI models. Our responsibilities are assessed according to our role, the technology used and the intended application of the final content.
The EU AI Act
Scenes supports the objectives and requirements of the EU AI Act: responsible innovation, human oversight, transparency and protection against manipulation and deception.
The transparency obligations in Article 50 have applied since 2 August 2026. They distinguish between providers of AI systems and professional users of those systems. For creative production, relevant issues include clear disclosure of deepfakes and editorial responsibility for AI-assisted publications.
We maintain appropriate AI skills within our team and continuously update our knowledge as technology, legislation and official guidance develop. We also help our clients translate the regulations into practical decisions about production, documentation, labelling and publication.
Transparency and labelling
The AI Act does not require the same label on every piece of content in which AI has played a role. The appropriate level of disclosure depends on how substantially AI has generated or manipulated the content, whether it could be mistaken for authentic documentation, and where and how it is published.
Scenes follows a transparency-first approach. When AI has materially contributed to an image or video, we classify the content as one of the following:
Fully AI-generated
Contains AI-generated elements or is partially AI-modified
AI-assisted editing, where AI has only performed standard supporting or editing functions
For each relevant delivery, Scenes can recommend appropriate disclosure wording and placement based on the content, channel and intended use.
Where visible disclosure is required, it must be clear, accessible and available when the audience first encounters the content. File metadata and Content Credentials can complement that disclosure; they do not replace a required visible or audible label. Evidently creative works may use appropriate disclosure that does not hamper their display or enjoyment.
Scenes supports the EU icons for labelling AI-generated content and can help clients determine when labels such as “Fully AI-generated” or “Partially AI-modified” are appropriate.
Scenes remains responsible for its own use and publication of AI content. Clients are responsible for their own publication decisions and context. Required disclosures should remain available on each channel, including separately shared and downloaded versions; an overlay on this website does not travel automatically with an exported file.
Disclosure on this website
Our portfolio contains AI-generated and AI-modified images and films, including fictional people, objects, rooms, locations and events. The creative-work context matters; disclosure is not limited to images containing faces.
We use the AI-generated mark for generated imagery and identify known composites as AI-modified or as containing AI-generated elements. Where a photograph of a product or person is combined with an AI-created setting, the original photograph remains part of the work. Original photographs of Michael Rying and source photographs identified in case studies are not presented as AI-generated.
The EU icons are optional tools for disclosure. Using an icon does not by itself establish legal compliance, and it does not imply that Scenes is a signatory to the voluntary Code of Practice.
Editorial responsibility
Michael Rying is Editor in Chief of As Scenes sees it. New AI-assisted articles require his substantive review and approval before publication. This includes reviewing the argument, factual claims and sources, rather than only checking spelling or grammar. Scenes holds responsibility for its own publication. AI assistance does not replace editorial judgment.
Truthful marketing and product representation
AI-generated environments can look entirely authentic. We therefore take particular care to ensure that commercial content does not mislead consumers about a product, service, location, person or event.
AI must not be used to fabricate or misrepresent material product characteristics, certifications, performance, scale, safety features or customer experiences. Product-specific claims and representations must be supported by information supplied or approved by the client.
Fictional environments must not be presented as documentary evidence of a real location, event or situation where this could materially mislead the audience.
Human oversight and quality control
Scenes does not treat unreviewed AI output as finished professional content.
Every final asset undergoes human creative direction, evaluation and quality control. Depending on the project, this may include checks of product accuracy, proportions, materials, perspective, lighting, anatomy, brand consistency and potentially misleading details.
We also consider the risk of bias, stereotypes, discriminatory representation and unrealistic portrayals of people. When people are depicted, we aim for respectful, credible and inclusive representation.
Intellectual property, identity and consent
Scenes uses professional AI services that permit the intended commercial use and complies with the applicable terms of those services.
We do not knowingly reproduce protected artworks, designs, photographs, trademarks or other third-party intellectual property without the necessary authorisation. Creative references may be used to communicate a direction, atmosphere or visual principle, but not to deliberately copy a protected work.
Scenes does not create unauthorised representations of identifiable private individuals, public figures or politicians, or use a person’s face, voice or identity in a misleading context. Written permission will be obtained where consent or other rights clearance is required.
The same principle applies to recognisable properties, artworks, architecture, products, logos and brands where their inclusion requires permission.
Privacy and confidential information
Personal data, confidential client information and unreleased commercial material must not be entered into an AI system unless that system has been assessed and approved for the intended use.
Client assets are processed in accordance with the relevant agreement and must not be used for unrelated purposes. We take reasonable precautions to protect confidential briefs, product information, personal data and other sensitive material throughout production.
This approach follows recommendations from the Danish Agency for Digital Government and the Danish Data Protection Agency.
Content provenance
Scenes ApS has been a member of the Coalition for Content Provenance and Authenticity—C2PA since 29 May 2024.
C2PA Content Credentials can provide machine-readable information about the origin and editing history of digital content. Where supported by the production tools and delivery format, Scenes supports the preservation and use of this information.
We also support technologies such as Google DeepMind’s SynthID, which can add machine-readable provenance information to content created with supported Google AI tools.
Our production workflow should preserve available provider-added provenance information, watermarks and machine-readable AI markings. Resizing, editing and format conversion can affect that information, so provenance needs to be checked for each final delivery. An AI label on this website does not certify that the media file contains valid Content Credentials.
Commercial usage rights
Commercial usage rights are defined in the individual project agreement. Unless otherwise agreed, the client receives the agreed commercial rights to use the final delivered assets.
Scenes cannot transfer rights that it does not own or guarantee that exclusively AI-generated elements qualify for copyright protection in every jurisdiction.
If an asset is subsequently altered, manipulated, combined with other material or used in a substantially different context, the client should reassess its accuracy, rights clearance and disclosure requirements. Scenes cannot remain responsible for changes made after final delivery.
Helping clients implement the rules
Scenes helps clients apply responsible AI principles in practice. Depending on the project, this may include:
Assessing how the AI Act may apply to the content
Classifying the level of AI involvement
Recommending suitable wording or EU labelling icons
Advising on visible disclosure, captions and metadata
Preserving available provenance information
Establishing documentation and approval procedures
Adapting the approach to different channels, markets and publication contexts
For regulated, politically sensitive, high-risk or legally complex applications, we recommend that the client also involves its own legal advisers.
Continuous review
AI regulation and technical standards continue to evolve. Scenes reviews this Code of Conduct at least annually and whenever significant new legislation, official guidance or provenance standards are introduced.
Relevant guidance includes the Danish Agency for Digital Government’s AI Act information, the EU guidelines on Article 50 and the voluntary EU Code of Practice on Transparency of AI-generated Content.
This Code of Conduct describes Scenes’ professional principles and working practices. It does not replace a project-specific legal assessment.