AI systems and purposes
Purpeu calls the OpenAI API for text generation, content analysis, image generation, and text-to-speech. The models currently configured are gpt-4o-mini for text and analysis, gpt-4o-mini for assistant replies,gpt-image-1 for images, and gpt-4o-mini-tts for speech. A configured model can change for security, quality, availability, or cost. Material changes will be reflected in this notice.
- Answering writing and marketing questions with selected brand and project context, and proposing content revisions for users to review before applying.
- Rewriting source material into social posts, carousels, scripts, and lead magnets.
- Analysing customer-provided visual references and layout instructions.
- Generating draft images and spoken audio at the customer's request.
Data sent to the provider
Depending on the feature, Purpeu sends prompts, selected source text, public webpage text imported at the customer's request, brand context, output settings, and uploaded reference material. Account passwords, payment card data, and unrelated team data are not included in AI prompts. Customers must avoid unnecessary personal data, secrets, special-category data, criminal-offence data, and children's data.
How the assistant works
You are chatting with an AI system. The assistant receives your message, selected brand details, relevant project material, and recent conversation. Long sources and older messages may be shortened, and the interface tells you when that happens. The assistant does not browse the web or publish content automatically.
Replies and proposals are saved in your private conversation. Only revisions you choose to apply change shared project content. Applying or undoing a revision clears previous client approval; affected media must be regenerated through the existing controls. Team reply allowances and other feature allowances remain separate.
Assistant requests disable provider response storage. Purpeu stores the conversation locally, while OpenAI's applicable abuse-monitoring and legal retention rules may still apply. The Privacy Policy explains access and deletion.
Limits and human review
A user decides what to generate, can edit the result, and must review accuracy, permissions, audience suitability, advertising claims, and disclosure duties before publishing. Purpeu does not use AI output to make decisions with legal or similarly significant effects about people.
Labels and provenance
Product screens and exported output identify generated material as AI-assisted or AI-generated where the format allows it. Purpeu preserves provider provenance metadata when supplied and adds machine-readable metadata to supported output formats where technically possible. File conversion, screenshots, copying, or a third-party platform can remove metadata, so visible disclosure may still be needed.
Users must clearly disclose realistic synthetic audio, image, or video that could be mistaken for a real person or event. Deepfakes and AI-generated text published to inform the public about matters of public interest require a clear, accessible disclosure unless a legal exception applies.
Questions and complaints
Questions about AI processing, output labels, personal data, or an unsafe result can be sent to [email protected]. Privacy rights and regulator contacts are explained in the Privacy Policy. Marketplace output can also be reported through the illegal-content form.