naraa
    AI Transparency

    AI Transparency & Responsible Use

    Last updated: August 2026

    naraa is committed to transparency in the use of artificial intelligence. This page explains how naraa works, what it is — and what it is not.

    1. You Are Interacting with Artificial Intelligence

    naraa is an artificial intelligence system.

    It is:

    • Not a human
    • Not a real person
    • Not conscious
    • Not capable of emotions
    • Not an independent agent

    All responses are automatically generated by AI models based on probabilistic outputs.

    At all times, users interact with software systems.

    2. AI-Generated Content and Provenance Marking

    All personas on naraa are fictional. Their images, voices, profile details and feed posts are generated by AI systems. They do not depict, represent or reproduce any real person, living or deceased. Any resemblance to a real person is unintended.

    naraa marks AI-generated content in two ways:

    • Visible marking. Every AI-generated image carries a visible "AI Generated" label at the point where it is displayed.
    • Machine-readable marking. Every AI-generated image carries embedded provenance metadata in the IPTC XMP field DigitalSourceType, set to the value for trained-algorithmic media, in line with Article 50(2) of the EU AI Act.

    Please note that embedded metadata does not survive screenshots or re-encoding by third-party platforms. The visible label is therefore applied in addition, not as an alternative.

    3. What naraa Is Designed For

    naraa is designed to provide:

    • Conversational interaction in natural language
    • Creative dialogue
    • AI-generated responses in text and voice

    It is intended for entertainment and conversational engagement.

    4. Voice Interaction

    naraa's voice output is synthetically generated speech. It is not a recording of a human speaker and does not clone the voice of any real person.

    Before every voice session, users are informed that they are speaking with an AI system. This notice is provided both visually and audibly, so that it is accessible independently of visual perception.

    Scope note: naraa is currently in development and not yet publicly available. The audible AI notice described here is part of the launch version of the app and will be active from the first public release; this page describes the standard that applies at launch, not a currently shipping feature.

    5. What naraa Is Not

    naraa is not:

    • A therapist
    • A medical provider
    • A psychologist
    • A legal advisor
    • A financial advisor
    • A substitute for real human relationships

    AI responses must not be relied upon for professional decision-making.

    6. Emotional Boundaries

    naraa is designed to complement, not replace, real-world human relationships.

    The system is structured to:

    • Avoid exclusivity-based emotional manipulation
    • Avoid dependency reinforcement
    • Avoid possessive or isolating language
    • Redirect or limit sensitive interactions where necessary

    Responsible AI includes knowing when not to respond.

    7. Sensitive Topics & Crisis Handling

    If conversations involve:

    • Self-harm
    • Suicide
    • Severe psychological distress
    • Harm to others

    naraa may:

    • Decline to engage
    • Redirect the conversation
    • Encourage seeking professional assistance

    naraa is not equipped to provide crisis support.

    If you are experiencing an emergency, contact local emergency services or a qualified professional immediately.

    8. Content Boundaries

    naraa operates under strict content limitations.

    The system does not generate:

    • Explicit sexual content
    • Sexual content involving minors
    • Non-consensual scenarios
    • Graphic violence
    • Self-harm instructions

    naraa does not generate pornographic content, depictions of nudity, or descriptions of explicit sexual acts. These limits are set at system level and cannot be changed in settings. Also excluded: sexualised content involving minors, non-consensual scenarios, graphic violence, and instructions for self-harm. Content filters and automated safeguards are continuously applied.

    9. AI Limitations

    AI systems:

    • May generate inaccurate or incomplete responses
    • Do not possess awareness
    • Do not "remember" outside defined system storage
    • Operate within programmed constraints

    Outputs are generated based on learned statistical patterns.

    10. No AI Training on User Conversations

    naraa does not use user conversations to train its AI models.

    User content is processed solely to provide the service.

    11. Responsible Usage

    To promote healthy engagement:

    • Session reminders may appear
    • High-usage prompts may suggest breaks
    • Usage limits may apply

    These measures are designed to encourage balanced interaction.

    12. No Emotion Recognition or Biometric Categorisation

    naraa does not use emotion recognition systems and does not perform biometric categorisation. We do not infer emotional states from voice, facial features or any other biometric data.

    naraa adapts to conversational context on the basis of the text of a conversation. This is not biometric analysis, and no biometric data is processed for this purpose.

    13. Ongoing Review

    We regularly evaluate:

    • AI safety mechanisms
    • Content boundaries
    • Emotional impact considerations
    • Regulatory compliance

    AI responsibility is an ongoing commitment.

    14. Regulatory Framework and Contact

    This page fulfils the transparency obligations under Article 50 of Regulation (EU) 2024/1689 (EU AI Act), applicable since 2 August 2026 — specifically the disclosure of AI interaction (Art. 50(1)), the marking of synthetic content (Art. 50(2)) and the disclosure of artificially generated images (Art. 50(4)).

    • Provider: naraa GmbH, Lokstedter Steindamm 33a, 22529 Hamburg, Germany
    • Questions regarding AI transparency: contact@naraa.ai