We only pass the input question to OpenAI (third-party). Everything is fully anonymized for openAI (no user or corp identification). We do not pass the user
feedback to OpenAI.
Data is stored the same way in the getAbstract database as all other information we currently have. It will not be any different than reading a summary on the
webpage or using the standard search functionality.
OpenAI does not use anything we pass on to them to train their model. Data is kept with a retention of 30 days from their side. (This is what they state on their website.)
We give the opportunity to opt-out/in from this Q/A tool at the corporation level.
Data FAQ
What happens with my company's data?
A user asks a question, this question is passed to openAI to create an answer
based on getAbstract resources. Nothing else is passed to OpenAI, only the
question text (No information about the corporation, user, not even the
anonymized ID number)
It is the policy of OpenAI not to use this data to train their model. They have a
retention policy of 30 days. (more information on their website
https://openai.com/policies/privacy-policy )
For the Q&A we use the same data privacy rules as with all other tools/features.
The question and the answer are stored and associated with the anonymous
unique user identifier.
Is data privacy a problem? Who gets to see these answers?
Only getAbstract see the answer, we will not share externally. Storing and analysing this data helps us improve the service. The stored data is anonymized.
Are the answers user-related and trackable?
Yes, we can associate a question-answer with the users' anonymized ID.
Is data being used anonymized?
Yes, at the same level as everything else on our website.
Do we share the answers outside of getAbstract?
No, we do not share the answers with anyone outside of getAbstract.
What if a user enters proprietary information in the question?
If users put proprietary information into the system, we can manually delete it from our side. There is no automatic detection.
What data does the tool store about an individual user e.g., are the historic
questions (and responses) stored, and if so for how long? If this is the case can this
data be linked to an individual user?
We plan in the future to allow the users to see their own history. So, we do store the history and we can associate it to the anonymized user id.
We have no limit in data retention. It is possible to ensure the data remains anonymized for users.
To avoid any data issues i.e. being linked to individual users, can we request that the questions/responses are deleted at the end of the session?
Right now, we have no way of deleting the usage created during the
session. With the design of anonymized data at the core of this tool, we feel it is
a significant opportunity lost not to have such valuable insights. It is a key
strategic benefit to getAbstract customers to understand important trends and
insights and to ensure that relevant knowledge and resources are aligned with
what people need.
