AI Usage Policy

Introduction

This AI Usage Policy explains how WebVoice uses Artificial Intelligence systems in compliance with:

• Regulation (EU) 2024/1689 - AI Act
• Italian Legislative Decree 196/2003 (Privacy Code)
• Regulation (EU) 2016/679 (GDPR)
• Guidelines from the Italian Data Protection Authority (Garante per la protezione dei dati personali)

1. AI Systems Used

1.1 Text-to-Speech (TTS)

• Model: Kokoro ONNX
• Type: Generative AI for speech synthesis
• Risk Level: Limited Risk (AI Act)
• Purpose: Convert text to natural-sounding speech

1.2 Speech-to-Text (STT)

• Model: Faster Whisper (based on OpenAI Whisper)
• Type: AI for transcription and language recognition
• Risk Level: Limited Risk (AI Act)
• Purpose: Transcribe audio to text with language detection

2. Transparency and Disclosure

In accordance with Article 13 of the AI Act, we inform you that:

• You are interacting with an AI system
• The AI system is used for TTS and STT processing
• AI-generated content may contain inaccuracies
• You should verify AI-generated transcriptions
• AI-generated speech is synthetic and may not perfectly match human speech

3. Data Processing

3.1 Input Data

For TTS:

• Text content you provide
• Language and voice preferences
• Speed parameters

For STT:

• Audio files you upload
• Optional language specification

3.2 Processing Location

All AI processing occurs on our servers within the European Union. Your data is NOT sent to external AI services or third-party providers.

3.3 Data Storage

• Input data: Processed in real-time, not permanently stored
• Output data: Stored only if you explicitly save it
• Anonymized data: May be used for model improvement

3.4 Service Limits

To ensure optimal performance and fair resource allocation, the following technical limits apply:

• Free Tier:
  • STT: Maximum audio file size 50.0 MB
  • TTS: Maximum text length 5000 characters
• Premium Tier:
  • STT: Maximum audio file size 500.0 MB
  • TTS (local Kokoro): Maximum text length 50000 characters
• Premium Subscription: Monthly subscribers are also subject to usage caps (video generations per day, cloud TTS rate limits, no free chat models, credit rollover cap). See Terms of Service for details.

These limits are enforced to prevent system abuse, ensure service availability, and maintain processing quality. Files or text exceeding these limits will be rejected before AI processing begins.

4. Accuracy and Limitations

AI systems have limitations:

• TTS may mispronounce uncommon words or names
• STT accuracy depends on audio quality and language
• Background noise may affect transcription accuracy
• Accents and dialects may not be perfectly recognized
• Technical or specialized terminology may be transcribed incorrectly

Important: Always review and verify AI-generated content before using it for critical purposes.

5. Human Oversight

In accordance with Italian AI guidelines:

• Users maintain full control over input and output
• You can review, edit, or delete generated content
• No automated decision-making affects your account
• Human review is available for support requests

6. Bias and Fairness

We acknowledge that AI systems may exhibit biases:

• Language models may favor certain dialects or accents
• Voice models may have gender or cultural biases
• We continuously work to improve fairness and reduce bias
• We provide multiple voice options for different languages

If you encounter bias or discrimination, please report it to: ai-ethics@webvoice.com

7. Prohibited Uses

You may NOT use our AI systems to:

• Generate deepfakes or impersonate individuals
• Create misleading or fraudulent content
• Violate copyright or intellectual property rights
• Generate harmful, illegal, or discriminatory content
• Bypass security measures or authentication systems
• Create content that violates privacy rights

8. Intellectual Property

AI-generated content:

• Belongs to you, the user who requested it
• May be subject to copyright of the original input
• Should not infringe on third-party rights
• You are responsible for ensuring compliance with applicable laws

9. Model Training and Improvement

We may use anonymized data to improve our AI models:

• Data is anonymized before use
• No personal identifiers are included
• You can opt-out by contacting us
• This helps improve accuracy and reduce bias

10. Right to Explanation

Under GDPR Article 22 and AI Act, you have the right to:

• Understand how AI decisions are made
• Request explanation of AI processing
• Challenge automated decisions
• Request human review

To exercise these rights, contact: ai-transparency@webvoice.com

11. Compliance with Italian Regulations

WebVoice complies with:

• Garante per la protezione dei dati personali guidelines on AI
• Italian AI Strategy requirements
• National guidelines on AI ethics
• Sector-specific regulations where applicable

12. Risk Assessment

Our AI systems are classified as 'Limited Risk' under the AI Act:

• No high-risk applications
• No automated decision-making affecting individuals
• Transparency requirements met
• Human oversight maintained

13. Incident Reporting

If you experience issues with AI processing:

• Report technical errors: support@webvoice.com
• Report bias or discrimination: ai-ethics@webvoice.com
• Report privacy concerns: privacy@webvoice.com

14. Updates to AI Systems

We may update AI models to improve performance. Significant changes will be communicated. Model versions are tracked for transparency.

15. Contact

For AI-related inquiries:

• General: ai-info@webvoice.com
• Ethics: ai-ethics@webvoice.com
• Transparency: ai-transparency@webvoice.com