Indeemo / Microsoft Azure Generative AI information

 

You have the option to upgrade your Indeemo subscription to activate Microsoft Azure’s Generative AI (which is based on the GPT-3.5 and GPT-4 models).

Microsoft is an existing sub-processor of Indeemo as stated in our Standard Terms of Service.

Although the Large Language Models (LLMs) used by Microsoft to provide the Generative AI service are based on the OpenAI GPT models, all of your prompts and data are processed exclusively by Microsoft.

This Microsoft Data, Privacy, and Security extract confirms their processing details:

 
 

As Microsoft is an existing Indeemo sub-processor, activating Generative AI on Indeemo does not alter our sub-processors or our Terms & Conditions of Service in any way.

How it works:

When the Generative AI capability is optionally activated on the platform, Researchers can use prompts to analyse text based data.

Analysis use cases typically include summarisation of text, identification of themes / topics in text, sentiment analysis, translation and tabulation of answers in video surveys. When a prompt is sent, the following text based data is processed by Microsoft:

  • Respondent first name (or pseudonym if used)

  • The following text based metadata: video transcripts, captions of photos / videos, notes or OCR text (text extracted from videos / screen recordings using Optical Character Recognition).

Here is an example of a typical prompt and result:

 
 
 

The key points to note are:

  • Use of the Indeemo/Microsoft Generative AI service in no way impacts your ownership of all IP in the Research Content. Indeemo, Microsoft do not make any claim on the IP in any data you process through the Indeemo/Microsoft Generative AI feature.

  • Data submitted through Indeemo’s Microsoft Generative AI service can be processed in the EEA, USA/Canada or Japan. As a default, Indeemo will process North American projects in the US / Canada region, Asia Pacific projects in Japan and all other Clients’ data will be processed in the EEA region. Any required changes to this can be discussed and configured at project set up.

  • All other Indeemo Services process data within Europe.

  • When data is submitted to the service, it is processed through content filters as well as the specified Generative AI model. The content filtering models are run on both the prompt inputs as well as the generated completions.

  • In addition to synchronous content filtering, the Microsoft Generative AI Service stores prompts and completions from the service for up to thirty (30) days to monitor for content and/or behaviours that suggest use of the service in a manner that may violate applicable product terms.

  • If the system identifies harmful content, customers receive either an error on the API call if the prompt was deemed inappropriate or the finish_reason on the response will be content_filter to signify that some of the generation was filtered.

  • No Prompts or Completions / Results are stored in the model during these operations. Furthermore Prompts and completions are not used to train, retrain or improve the models.

For wider context, the Data, Privacy, and Security details for the Microsoft Generative AI Service is available at this link.

If you have any questions on the above, please contact dataprotection@indeemo.com

 
information security certifications.png
 

Finally:

  • If you have any questions pertaining to Privacy and Security, please do not hesitate to Contact Us.