RESPONSIBLE AI POLICY

1. Introduction

Orbitum Tech OÜ (Registration Number: 17390885) operates sec30.com ("the Service"), an AI-powered creative platform. This Responsible AI Policy sets out the ethical framework that governs how we develop and operate AI features, and how we are transparent with you about the AI-generated nature, data handling, and limitations of the technology. It should be read together with our Terms and Conditions, Privacy Policy, and Acceptable Use Policy.

2. Our Responsible AI Principles

We are committed to the responsible development and use of artificial intelligence. Our work is guided by the following principles:

• Lawful and safe by design. We build safeguards into the Service to prevent the generation of harmful, illegal, or unethical content.
• Transparency. You should always know when you are interacting with AI, that outputs are machine-generated, and what their limitations are.
• Human accountability and oversight. AI does not make consequential decisions about users on its own. People design our safeguards, monitor the system, and make the decisions that affect accounts.
• Privacy and data minimisation. We do not use your content to train AI models, and we share your inputs with providers only as needed to generate the output you requested.
• Fairness and awareness of bias. We recognise that AI models can reflect bias present in their training data, and we do not endorse AI outputs. We act on credible reports of harmful or discriminatory behaviour.
• User control and responsibility. You direct what the Service produces, and you remain responsible for reviewing and validating outputs before you rely on or distribute them.
• Provider due diligence. We access AI models only through their providers' official, authorised APIs, and we require compliance with the providers' own policies.

3. Transparency & AI Disclosure

You are interacting with AI

The chat responses and images produced by the Service are generated by automated artificial intelligence models based on your inputs. They are not written or reviewed by a human, and you are not communicating with a person when you use these features.

Outputs can be inaccurate or fabricated

AI-generated outputs may be inaccurate, incomplete, misleading, biased, or entirely fabricated (a phenomenon commonly referred to as "hallucination"). Outputs do not constitute professional, legal, medical, or financial advice. You are solely responsible for reviewing, verifying, and validating any output before relying on or distributing it. These limitations are also described in Section 2A of our Terms and Conditions.

How your data is used

To generate outputs, your prompts and inputs are transmitted to third-party AI providers (OpenRouter and fal.ai) solely for the purpose of producing the content you request. We do not use your content to train AI models, and we do not permit these providers to use your content for training. A full description of what data we process, who we share it with, international transfers, and your rights is set out in our Privacy & Cookie Policy (Sections 4, 5, and 11).

4. Content Labelling

We identify AI-generated content within the Service so that it is not mistaken for human-authored material:

• AI chat responses are visually distinguished and labelled as produced by the AI assistant, separately from your own messages.
• Images produced by the Designer agent are identified as AI-generated within the application interface and gallery.

Please note that labelling applied within the Service may not persist once content is exported, downloaded, or shared outside the platform. Where you publish or distribute AI-generated content, you are responsible for disclosing its AI-generated nature where required by applicable law or platform rules.

5. Content Moderation & Safety

The Service applies automated, multi-layer content moderation to inputs before they reach any AI model, in order to prevent the generation of prohibited content such as child sexual abuse material, deepfakes and non-consensual imagery, explicit adult content, violence and self-harm, fraud, and infringing material. The categories we prohibit and the way we enforce them are described in full in our Acceptable Use Policy.

6. Human Oversight & Intervention

We operate a human-on-the-loop oversight model. AI is used to generate content and to pre-screen inputs, but it does not autonomously make consequential decisions about users. Specifically:

• Humans design the guardrails. Our staff define and maintain the moderation rules, classifier gates, model selection, and agent configuration that govern the Service.
• Humans monitor the system. Every automated moderation block is recorded in an audit log that is reviewed by our Trust & Safety function to detect errors, abuse, and patterns requiring intervention.
• Humans make the consequential decisions. Actions such as content removal, account suspension, account termination, and refunds are taken by a member of staff, not by an automated system.
• Humans can intervene. Our staff can update or disable moderation rules, change or disable models, remove content, and suspend accounts where necessary.

Reporting and appeals

If you believe an automated moderation decision was made in error, or you wish to contest an account action, or you wish to report a harmful or problematic AI output, you can request human review by emailing [email protected]. A member of our Trust & Safety function will review your case. We aim to acknowledge such requests within 2 business days and to provide a final decision within 15 business days.

7. Governance & Review

This policy is owned by the management of Orbitum Tech OÜ and is reviewed periodically, and whenever there is a material change to our AI providers, features, or applicable law. Our operational human oversight and intervention procedures are maintained internally and applied by our Trust & Safety function.

8. Contact

For questions about this Responsible AI Policy, or to request human review of an AI-driven decision or output, please contact: