AI Transparency Notice
Last updated: 1 November 2026 · Version 1.0
AutoFeed, provided by AG IT-Solutions ("AGITS"), uses artificial intelligence (AI) to help you create social media content. In the spirit of article 50 of the EU AI Act (Regulation (EU) 2024/1689), this notice explains where AI is used, which provider we rely on, what happens with your data and what remains your responsibility.
1. Where AutoFeed uses AI
| Feature | What the AI does | Output |
|---|---|---|
| Captions and hashtags | Writes or rewrites post texts and suggests hashtags based on your draft, brand voice and strategy | Text |
| Content calendar and strategy | Suggests post ideas, themes, timing and content strategy | Text, suggestions |
| Create from text, idea or link | Turns a text, idea or web page into draft posts | Text, post drafts |
| Templates | Suggests branded template designs and fills them with text | Template layouts, rendered images |
| Image generation | Creates new images from a description | Synthetic images |
| Image editing | Modifies existing images (e.g. background, style, elements) | Edited images |
| Media library | Describes and tags your images to make them searchable and reusable | Descriptions, tags |
| Brand import | Helps interpret your website's colours, fonts and logo | Brand settings |
| Reports | Summarises analytics and suggests improvements (paid Plans) | Text |
| Chat assistant | Answers questions and performs actions in your workspace at your request | Text, actions |
You always know when you interact with AI: AI features are clearly marked in the app, and the chat assistant is an AI system, not a human.
2. Our AI provider
- Model provider: Google LLC, via the Google Gemini API (paid tier) for text generation, image analysis and image generation and editing.
- Orchestration: AI requests are coordinated by AGITS Agentic, AGITS's own agent platform, running in the AWS EU region (Ireland).
- Google may process requests in the United States or elsewhere, under the EU-US Data Privacy Framework and Standard Contractual Clauses. See Sub-processors.
We may change or add AI providers; changes are announced through our sub-processor list.
3. Your data and model training
- We do not use your content to train AI models. Not our own, not anyone else's.
- Under Google's terms for the paid Gemini API, prompts and outputs are not used to train Google's models. Google may retain data for a limited period to detect abuse, as described in its terms.
- We send the AI only the context needed for the specific request (for example the draft you are editing, your brand settings or the image you want to edit).
- Do not include sensitive personal data in prompts unless necessary.
4. Limitations of AI
AI can be wrong. AI Output may:
- contain factual errors, outdated information or invented details ("hallucinations");
- be unsuitable for your audience, culture or language register;
- resemble content generated for others, or (rarely) resemble existing protected works;
- reflect biases present in the training data of the model;
- contain spelling errors or distorted elements, especially text inside generated images.
AI suggestions about strategy, timing or performance are indicative and do not guarantee results.
5. Your responsibility: review before publishing
- You remain in control. Under Settings → Automation you choose how much the AI does: Manual (AI only helps when asked), Co-pilot (AI plans and writes, you approve before anything is published) or Full auto (AI plans, writes and publishes on its own, on plans that include it). AI-made content is only published without your prior approval if you have explicitly chosen automatic publishing.
- Automatic publishing is your decision. If you enable it, you accept that posts go live without human review and you remain fully responsible for them. We recommend approval mode for regulated sectors, sensitive topics or campaigns with legal claims. You can switch back at any time.
- You are responsible for what you publish, including AI Output, as if you had created it yourself. Check facts, prices, claims, names and legal statements.
- Follow our Acceptable Use Policy, including the ban on deepfakes of real persons without their consent.
6. Labelling AI-generated images
- Images generated or edited with AI are marked as such in your media library.
- confirm whether generated images carry an invisible watermark/metadata (e.g. Google SynthID / C2PA) and whether AutoFeed adds a visible or metadata label
- When you publish realistic AI-generated or substantially AI-edited images (for example images that look like real photos of people, places or events), label them as AI-generated, for example using the "AI info" label on Instagram and Facebook or a note in your caption. Meta may also apply such a label automatically.
- Purely illustrative, clearly artistic or minor edits (such as colour correction or cropping) generally do not need a label, but transparency towards your audience is always good practice.
7. No automated decisions about you
AutoFeed does not use AI to make decisions that produce legal effects or similarly significantly affect you. AI is used only to assist with content creation.
8. Questions and feedback
Did the AI produce something harmful, offensive or infringing? Report it to hello@agits.be. Privacy questions: privacy@agits.be.
9. Language
This notice is available in English and Dutch. Both language versions are equally authoritative.
Related documents: Terms of Service · Privacy Policy · Acceptable Use Policy · Sub-processors