How AI uses my data
Your data does not train the models used by Dixit; some information is sent to the model provider while a request is being processed.
The question we are asked most in vendor review. Here is the detailed answer.
In detail
What is not used to train the models used by Dixit
We have obtained from each of our language model providers a contractual undertaking that your data is not used for training. This covers your keywords, searches, annotations, labels and comments.
What may be sent while a request is processed
To score alert relevance and produce personalised summaries, the context sent to the model provider may include:
- the keyword expression for the relevant alert,
- the alert name when it is useful to the task,
- your organisation name,
- the policy priorities you entered in your profile.
This context lets the model explain why a document matters to you instead of producing a generic summary.
What is not sent for this processing
Your comments and labels are not sent to the model provider to score alert relevance or write your daily summary.
Your personalised summaries
They remain visible only to you, including when annotation sharing is enabled within your organisation.