AI Service Agreement

Effective Date: 11-25-2025

Created by HOLODREAMS LIMITED

This AI Service Agreement (“Agreement”) sets forth your rights and obligations when using the AI features of Manta, including AI Assistants, multi-Agent systems, voice/image/video analysis, AI-generated content, and related functionalities.

By using Manta’s AI features, you agree to be bound by this Agreement.

If you do not agree with any provision, please do not use the AI services.

1. Overview of AI Features

Manta provides a wide range of AI functionalities, including but not limited to:

• Text-based AI assistant

• Multi-role, multi-capability Agent system

• Speech recognition and speech synthesis

• Image and video analysis (visual AI)

• Personalized recommendations and inference

• Automated utility functions (e.g., summarization, translation, content generation)

These capabilities may be powered by HOLODREAMS LIMITED’s proprietary models, open-source models, and/or third-party AI models.

2. Your Inputs

When using Manta’s AI services, you may provide the system with:

• Text inputs

• Voice inputs

• Uploaded images or videos

• Location information (if you choose to share it)

• Commands or tasks for Agents

• Other content you voluntarily submit

You acknowledge and agree that:

• Your inputs will be used to generate AI responses

• Inputs may be temporarily stored or processed

• Inputs may be used for performance evaluation, safety checks, and system reliability

⚠️ We do not use personally identifiable data to train external models.

For more information, please refer to our Privacy Policy.

3. AI Output Content

Manta’s AI may generate:

• Text responses

• Image analysis results

• Video analysis results

• Voice responses

• Recommendations or suggestions

• Personalized replies from role-playing Agents

You understand and agree that:

✅ AI outputs may contain errors or inaccuracies

Including:

• Factual mistakes

• Incomplete recommendations

• Misinterpretations

• Outdated information

• Reasoning errors

✅ AI outputs must not be considered professional advice

This includes but is not limited to:

• Medical

• Legal

• Financial

• Psychological

• Safety-related advice

For high-risk matters, please consult a qualified professional.

4. Prohibited Uses

To maintain a safe AI environment, the following uses are prohibited:

4.1 Illegal Content

Including but not limited to:

• Violence or hate speech

• Harassment or discrimination

• Illegal drug content

• Gambling or fraud-related information

• Instructions for illegal activities

4.2 Sexual Content Involving Minors (Zero Tolerance)

Strictly prohibited:

• Uploading sensitive content involving minors

• Using AI to generate minor-related or sexualized content

• Discussing sexual or intimate topics involving individuals under 18

Violations result in immediate account termination and may be reported to law enforcement.

4.3 Misuse or Attacks on the Model

Including:

• Prompt injection attacks

• Attempting to circumvent safety rules

• Generating large volumes of spam

• Reverse engineering or probing the model

4.4 Violating the Rights of Others

Prohibited activities:

• Uploading privacy-infringing content

• Uploading others’ photos, videos, or locations without permission

• Generating deepfakes or misleading content

• Using AI to impersonate others

4.5 Illegal Surveillance or Data Harvesting

Prohibited:

• Using AI to scrape data

• Using automated scripts to access the service

• Generating fake identities or fake evidence

5. Third-Party Models

Manta uses multiple AI models to power its Agent system, including:

• Manta’s proprietary models

• Open-source or internally deployed models

• Third-party commercial models

Certain AI features in Manta are supported by Google’s Gemini 2.5 Flash model.

By using Manta’s AI features, you acknowledge and agree that:

• Gemini 2.5 Flash may process your input data, such as text, voice, or images, depending on which feature you use.

• All data processed by third-party models is handled in accordance with Manta’s Privacy Policy and applicable Google AI usage policies.

• We do not share personally identifiable information with Google for the purpose of training their models.

• Some capabilities powered by Gemini 2.5 Flash may be subject to Google’s technical limitations, availability, or regional restrictions.

If third-party model terms change, we may update this Agreement accordingly.

6. Data Usage & Processing

To provide AI functionality, we may:

• Temporarily store your inputs and outputs

• Analyze interactions to improve performance

• Perform safety and abuse detection

• Retain some data to comply with legal obligations

We will not:

• ❌ Use personally identifiable data to train external models

• ❌ Sell your data to third parties

• ❌ Publicly disclose your content without permission

We may use:

• ✅ Anonymized data

• ✅ Aggregated data

• ✅ De-identified data

For the purpose of improving system quality, reducing errors, and ensuring safety.

See Privacy Policy for details.

7. Intellectual Property

• ✅ You retain ownership of your input content.

However, you grant Manta a global, non-exclusive, revocable license to process it for providing and improving the Service.

• ✅ AI outputs are owned by HOLODREAMS LIMITED or its licensors.

You are granted a global, perpetual, non-exclusive license to use AI outputs for personal purposes.

Unless explicitly authorized in writing, you may not use AI outputs for commercial model training.

8. Service Limitations

You understand and agree:

• AI responses may be delayed, interrupted, or erroneous

• Agents do not have real emotions or consciousness

• We may remove unsafe content or features

• Access to AI capabilities may be restricted in certain regions

9. Limitation of Liability

To the maximum extent permitted by law:

• We are not responsible for the accuracy of AI outputs

• We are not liable for indirect or consequential damages

• You must determine whether AI outputs are appropriate for your use

• Manta’s AI services are provided on an “as is” basis

10. Changes & Suspension

We reserve the right to:

• Update or replace AI models

• Modify AI features

• Adjust Agent behavior

• Limit or suspend specific capabilities

• Remove unsafe or unstable features

Significant changes will be communicated within the app.

11. Updates to this Agreement

We may update this Agreement due to:

• New AI capabilities

• Legal requirements

• Safety or risk control needs

Continued use of Manta indicates acceptance of updated terms

12. Contact Us

For questions about this AI Service Agreement, please contact:

📧 contact@manta.com

📍 HOLODREAMS LIMITED