Are My Data Used for AI Training at Dubly?

No Use of Your Data for Model Training – Ever

Dubly does not train language models, voice models, or lip sync systems using customer data.
Your video content, voices, transcripts, or translations are never used to improve general AI systems.

‍

That means:

• No content is extracted to train future models
  
• No machine learning is applied to your material unless it directly serves your specific request
  
• Even features like voice cloning (e.g. using the original voice in other languages) are performed only for your use — never for training the system
  

Your data stays yours, and is never repurposed.

‍

‍

Your Own Secure Sandbox

All customer content is processed in a separate and isolated workspace — often referred to as a "sandbox."

‍

What that means:

• No data sharing between customers
  
• No access from other accounts
  
• No cross-project contamination
  

Even if you're using our API or submitting hundreds of videos in bulk, the data stays within your secure account context.

‍

‍

Automatic Deletion After Processing

At Dubly, data is retained only as long as necessary to fulfill your project.

• After translation and delivery, all source files and intermediate data are automatically deleted
  
• You can also manually delete content at any time
  
• All backups, logs, and infrastructure follow strict encryption and retention protocols, fully GDPR-aligned
  

You stay in control – not just during processing, but also after completion.

Full GDPR Compliance – Built in Europe

Dubly is a German company based in Leverkusen and Frankfurt.
We operate under strict European data protection standards — by design.

‍

That includes:

• No data processing on non-EU servers
  
• Subprocessors only used under EU Standard Contractual Clauses (SCCs)
  
• Optional DPA, NDA, and TOM documents available on request
  
• Full support for data access, deletion, and audit requirements
  

Our infrastructure is designed for privacy-first organizations.

‍

‍

{{cta}}

‍

{{callout}}

‍