AInVFX statement of AI ethics
Our approach to AI ethics ensures that technology empowers creativity while upholding UNESCO's core values for ethical AI, with concrete commitments on transparency, data provenance, creator consent, AI literacy, and frugal AI.
In no other field is the ethical compass more relevant than in artificial intelligence. These general-purpose technologies are re-shaping the way we work, interact, and live. The world is set to change at a pace not seen since the deployment of the printing press six centuries ago. AI technology brings major benefits in many areas, but without the ethical guardrails, it risks reproducing real world biases and discrimination, fueling divisions and threatening fundamental human rights and freedoms.
Gabriela Ramos, Assistant Director-General for Social and Human Sciences of UNESCO
At AInVFX, we believe AI should empower people, not replace them. Acknowledging AI's transformative impact on creativity and storytelling, we are committed to ensuring that artificial intelligence serves humanity in an ethical, fair, and sustainable way. Our framework aligns with UNESCO's recommendation on the ethics of artificial intelligence.
Core values for ethical AI
1. Human rights and human dignity
AI should respect, protect, and promote human rights and fundamental freedoms. We design AI solutions that enhance human creativity, not diminish it. Our technologies are rigorously tested to avoid discrimination, privacy violations, or social harm.
2. Living in peaceful, just, and interconnected societies
AI must contribute to a more just and inclusive world. We prioritize transparency and accountability, ensuring that our AI-driven tools do not amplify misinformation, divide communities, or undermine democratic processes.
3. Ensuring diversity and inclusiveness
AI should reflect the richness of human diversity. We actively mitigate bias by:
- Using diverse and representative datasets.
- Engaging with artists and researchers to ensure AI benefits everyone.
4. Environment and ecosystem flourishing
Sustainability is central to ethical AI. We minimize our environmental footprint by:
- Selecting and optimizing AI models for energy efficiency.
- Partnering with cloud providers committed to renewable energy.
Our commitment to AI ethics principles
We uphold UNESCO's ten guiding principles for ethical AI, ensuring our technologies are fair, accountable, and beneficial to society.
1. Proportionality and do no harm
We prioritize responsible AI use in creative fields to enhance human expression, avoiding harmful or misleading content. We focus on AI augmenting human creativity and artistic integrity, and avoid uses like deepfakes or discriminatory content.
2. Safety and security
We promote secure practices in AI workflows to protect creative assets and intellectual property. We educate on ethical and secure AI tool use, avoiding copyright infringement and misuse for harmful content.
3. Right to privacy and data protection
We advocate for responsible data handling in creative AI, advising on privacy best practices, anonymization, and respecting artists' data rights, while complying with relevant data protection regulations.
4. Multi-stakeholder and adaptive governance & collaboration
We actively engage with the creative community to foster dialogue on ethical AI in creative industries. We collaborate with stakeholders to develop best practices and adapt our approaches as AI evolves, encouraging community feedback.
5. Responsibility and accountability
We emphasize human accountability in creative AI. AI is a tool, and humans are responsible for creative outputs. We stress human oversight, and we advocate for documenting the human contribution to each AI-assisted asset, so authorship and chain of title stay clear.
6. Transparency and explainability
We demystify AI for creative professionals by making tools and workflows clear and accessible. We put open source and open weights first, because tools anyone can inspect are the foundation of transparency and accountability. When a hosted or closed tool is the better fit for a specific need, we say so and document the trade-offs, so people can make well-informed decisions about AI's true capabilities and limitations.
7. Human oversight and determination
We champion "human-in-the-loop" creative AI, ensuring AI empowers human creativity, not replaces it. Our services ensure artists remain in control, using AI to realize their artistic visions.
8. Sustainability
We promote AI use that supports the long-term viability of creative careers and the cultural landscape. We encourage awareness of AI's energy consumption and advocate for responsible resource use in creative workflows.
9. Awareness and literacy
We are dedicated to raising AI awareness and literacy within the creative community through our training programs. We empower creative professionals to ethically utilize AI tools.
10. Fairness and non-discrimination
We promote fairness and non-discrimination in creative AI, addressing potential biases in AI tools and datasets. We advocate for AI use that fosters diversity, inclusivity, and accessibility in creative content and workflows.
What this means in practice
Principles need concrete commitments. Here are ours, as of 2026:
- We label generated content. The EU AI Act, article 50, applicable since August 2, 2026, requires AI-generated content to carry machine-readable marking and to be disclosed to the public. We disclose when content we publish is AI-generated, we are adopting C2PA metadata on our outputs (a standard that records in the file, in a signed way, who created the content and with which tool), and we teach this marking workflow in our trainings.
- We build AI literacy. The EU AI Act, article 4, applicable since February 2, 2025, requires documented AI training for people who operate AI systems. Our documented training programs help studios and schools meet this obligation.
- We check training-data provenance. Before recommending or teaching a model, we assess where its training data comes from. We put models with documented or licensed corpora first.
- We ask creators first. We never train or fine-tune a model on a creator's work without their written permission, and we advise our clients and participants to hold the same line.
- We keep AI frugal. We size models to the task, favor local and quantized models where quality allows, and use the AFNOR SPEC 2314 frugal-AI referential to measure and reduce the environmental impact of AI workloads.