Ethical Framework

1. Purpose

This Ethical Framework sets out the principles and rules that govern how Aivon Vision ("Aivon", "we", "us") develops, operates, and continuously improves its AI image generation services. It applies to Aivon as a provider and, by extension through our Terms of Service, to every user of the platform.

We recognise that AI-generated content carries societal responsibilities. This document is our public commitment to operating responsibly.

2. Core Principles

2.1 Human Dignity

We do not permit the creation of content that degrades, dehumanises, or harms individuals or groups on the basis of race, ethnicity, gender, sexual orientation, religion, disability, or any other protected characteristic. Content designed to demean or harass specific individuals is strictly prohibited.

2.2 Honesty and Transparency

All images generated by Aivon are labelled as AI-generated through embedded metadata. We do not misrepresent the nature of our outputs. We are transparent with users about what our AI models can and cannot do, and about the limitations of the service.

2.3 Safety by Design

Our platform is designed to prevent harmful outputs at every layer - through model selection, prompt filtering, output review, and user reporting mechanisms. Safety is not an afterthought; it is a design requirement.

2.4 Fairness and Non-Discrimination

We acknowledge that AI models can reflect and amplify biases present in their training data. We actively monitor for discriminatory outputs and take steps to mitigate bias. We do not build or operate systems designed to discriminate against individuals or groups.

2.5 Privacy

We do not use images uploaded by users to train or fine-tune AI models without explicit written consent. User-generated content is processed solely for the purpose of delivering the requested service. Our Privacy Policy governs all data handling.

2.6 Environmental Responsibility

AI inference is computationally intensive. We are conscious of our energy footprint and seek to use efficient models and infrastructure. We will publish energy usage information as our operations scale.

3. Prohibited Use Cases

The following use cases are strictly prohibited on Aivon regardless of the stated intent of the user:

• Generation of child sexual abuse material (CSAM) or any sexualised content involving minors
• Creation of non-consensual intimate imagery (NCII) or deepfake pornography of real individuals
• Generation of content designed to harass, threaten, or stalk a specific individual
• Creation of disinformation, fake news imagery, or content designed to manipulate public opinion through deception
• Synthesis of imagery depicting real individuals in illegal, defamatory, or reputationally damaging scenarios without their consent
• Generation of content that glorifies, promotes, or instructs in violence, terrorism, or self-harm
• Creation of content that infringes copyright, trademarks, or other intellectual property rights
• Any use that violates applicable law in the user's or our jurisdiction

Violation of any of the above will result in immediate account suspension and, where legally required, reporting to relevant authorities.

4. Responsible AI Procurement

We evaluate the AI models we deploy against the following criteria before integration:

• Publicly available model cards or technical documentation
• Evidence of safety evaluations or red-teaming by the model developer
• Licence terms compatible with commercial use and ethical operation
• Absence of known critical safety failures or documented misuse at scale

We currently use FLUX.1-schnell by Black Forest Labs via the HuggingFace Inference API. We will update this document when our model stack changes.

5. User Responsibilities

Users of Aivon agree, through our Terms of Service, to:

• Use the platform only for lawful purposes
• Not attempt to circumvent safety filters or prompt restrictions
• Not represent AI-generated images as authentic photographs of real events without disclosure
• Report content they believe violates this framework via our Contact page

6. Enforcement

Aivon enforces this framework through a combination of automated filtering, user reporting, and human review. Accounts found in violation will receive a warning for minor first offences or immediate suspension for serious violations. There is no appeal for CSAM-related violations.

Our full enforcement process is described in the Human Oversight Process.

7. Review and Updates

This framework is reviewed at minimum every six months, and following any significant change to our technology stack, applicable law, or the broader AI regulatory environment. The current version and effective date are shown at the top of this page.

8. Contact

Questions or concerns about this framework can be directed to [email protected] or via our Contact page.