Auto Subtitle Transcription: Frequently Asked Questions
This FAQ addresses common security concerns about data residency, encryption, and the use of organizational data in machine learning training. Our primary goal is to ensure that while we leverage AI for efficiency, we maintain a "Security First" posture that aligns with our internal compliance frameworks.
What feature uses AI?
We have an Auto Subtitle feature for uploaded demo content, that the customer can choose to enable on a demo-by-demo basis. We are utilizing AWS Transcribe to transcribe the audio in demos into text so customers can have auto subtitles generated on their demos.
What AI models are used?
We are utilizing AWS Transcribe to transcribe the audio in demos into text, when the customer chooses to turn on auto subtitles.
What types of data will the AI access, process, and store?
The AI will process the audio from a demo into text. The types of data processed depend on what the speaker in the demo says during the demo. All transcriptions are stored in AWS for the life of the customer's contract with Consensus + 30 days, unless otherwise specified in the customer's agreement with Consensus.
How is data collected by the AI, and what are the sources?
The audio data is collected when you upload the demo to Consensus. The sources are the voice(s) of the speaker(s) during the demo.
What is the purpose of collecting and processing this data?
When the customer invokes this feature, the purpose of collecting and processing the audio from the uploaded demo is to generate auto subtitles for customers to use in their demo.
Is customer data used to train AI models?
No, this data is not used to train any AI models.
Is any PII collected? What safeguards does Consensus have to mitigate IP or copyright risk for this AI feature?
PII is not being collected explicitly in this instance, as Consensus is simply generating subtitles for the uploaded demo. The auto subtitles are generated from the uploaded demo content that you, as a Consensus customer, choose to upload to your instance of Consensus . Consensus requires all customers to prevent demo transcripts (as well as other customer data) from being infringing, so to the extent a generated response includes customer data the customer is responsible for any infringing material.