- Claude operates as a closed system, without connection to the public internet.
- The alt-text generation process is done in isolation, and the content is not retained or used for any other purpose.
- Publisher images and the generated alt text are not stored or used to train or improve the AI model.
- The system does not ingest or learn from content in any way that could compromise intellectual property.
Anthropic's policy:
... [Anthropic] may use Materials to provide, maintain, and improve the Services and to develop other products and services. We will not train our models on any Materials that are not publicly available, except in two circumstances:
- If you provide Feedback (through the Services or otherwise) regarding any Materials, we may use that Feedback in accordance with Section 5 (Feedback).
- If your Materials are flagged for trust and safety review, we may use or analyze those Materials to improve our ability to detect and enforce Acceptable Use Policy violations, including training models for use by our trust and safety team, consistent with Anthropic’s safety mission.
Accordingly, DPS provides these guarantees:
- Any feedback generated and shared through either DPS or the Claude AI system's feedback feature must absolutely be approved by the DPS client before it is submitted.
- No Feedback will be shared.
- If any DPS client image is flagged by Anthropic for trust/safety review, DPS will notify the the client.
We believe this solution offers a good balance of efficiency and security, allowing clients to leverage AI capabilities without risking exposure or misuse of publisher-owned content.