One of the most frequently asked questions about Mediagraph's suite of artificial intelligence features is:
What happens to our data? Will our images and documents be used for training AI models?
Mediagraph uses outside service providers for AI functionality. This allows us to provide services on the absolute cutting-edge of this constantly-changing technology.
AI services include:
- Auto-keywording, Auto-generated Alt Text and Custom AI Actions powered by Anthropic's Claude (via Amazon Bedrock)
- Facial recognition and Sensitive Content Warnings powered by Amazon Rekognition
- Transcription and OCR powered by Amazon Textract
Both Amazon and Anthropic (the Large Language Model provider) have clear policies about model training. Both state in their documentation that client data supplied via the commercial API is not retained or used for model training.
These AI features are only activated when explicitly enabled in your account settings. When active, only relevant asset data is processed (such as the files themselves for keywording or facial recognition, or extracted textual content for transcription and analysis). All customer data for facial recognition remains securely within Mediagraph's AWS environment and is never shared with external parties.
For complete details about these privacy protections, please refer to the official documentation from our service providers:
- Anthropic's data training policy
- Anthropic's Commercial Terms of Service
- Amazon Bedrock data protection documentation