AI Nutrition Fact Labels
This page covers the AI Nutrition Fact Labels for the Conversational Intelligence Base Models. These labels detail how Twilio handles customer data across the Base Models, with specific guidelines on data logging, privacy, and compliance.
Twilio uses different Base Models as the foundation for creating Transcripts and executing Language Operators. Conversational Intelligence includes the following AI Base Models:
- Twilio's own AI Model
- OpenAI
- Amazon Comprehend
- Deepgram Speech-to-Text
- Google Speech-to-Text
- Amazon Transcribe
- Private AI
Base Model training
Customer data is not used to train the Base Models for OpenAI, Deepgram Speech-to-Text, Private AI, Google Speech-to-Text, or Amazon Transcribe.
Twilio uses customer data to train the Twilio Base Model only if you opt in by accepting the Data Logging Terms when you create your Intelligence Service. For more information, refer to the Twilio Conversational Intelligence: Data Logging Consent Addendum.
AI Nutrition Fact Labels give consumers and businesses a more transparent and clear view into "what's in the box" of a specific AI model. The Label explains how the AI model uses customer data for training AI models and how it handles data privacy and compliance.
The following sections show the current AI Nutrition Fact Labels for the Conversational Intelligence Base Models. Visit Twilio AI Nutrition Facts to learn more about AI Nutrition Facts and how to read these labels.
AI Nutrition Facts
Conversational Intelligence - Twilio models
- Description
- Twilio Automatic Speech Recognition that creates transcripts and Pre-built Language Operators that generate structured information mapping to specific data types such as agent introduction or voicemail detection.
- Privacy Ladder Level
- 4
- Feature is Optional
- Yes
- Model Type
- Predictive
- Base Model
- Twilio
- Base Model Trained with Customer Data
- Yes
- Customer Data is Shared with Model Vendor
- Yes
- Training Data Anonymized
- No
- Data Deletion
- Yes
- Human in the Loop
- Yes
- Data Retention
- Until the customer deletes
- Logging & Auditing
- Yes
- Guardrails
- Yes
- Input/Output Consistency
- Yes
- Other Resources
- https://www.twilio.com/docs/conversational-intelligence/pre-built-operators
Trust Ingredients
If the customer enables Data Logging functionality, then the Base Model is trained with customer data.
If the customer enables Data Logging functionality, then customer data is shared with Twilio to train the Base Model.
If the customer enables Data Logging functionality, then trained data might be deleted by the customer through written request. The customer can delete non-training data at any time using the API.
The customer views output in the Conversational Intelligence API or Transcript Viewer.
Compliance
Learn more about this label at nutrition-facts.ai
AI Nutrition Facts
Conversational Intelligence - OpenAI
- Description
- Language Operators with OpenAI in Conversational Intelligence. Used for Generative Custom Operators, as well as Prebuilt Operators for Summarization and Sentiment Analysis.
- Privacy Ladder Level
- N/A
- Feature is Optional
- Yes
- Model Type
- Generative
- Base Model
- OpenAI GPT-4o mini
- Base Model Trained with Customer Data
- No
- Customer Data is Shared with Model Vendor
- No
- Training Data Anonymized
- N/A
- Data Deletion
- Yes
- Human in the Loop
- Yes
- Data Retention
- Until the customer deletes it
- Logging & Auditing
- Yes
- Guardrails
- Yes
- Input/Output Consistency
- Yes
- Other Resources
- Generative Custom Operators: https://www.twilio.com/docs/conversational-intelligence/generative-custom-operators, Prebuilt Operators: https://www.twilio.com/docs/conversational-intelligence/pre-built-operators
Trust Ingredients
Customer data is shared with the Model Vendor for processing only. Customer data isn't used to train the Base Model.
Language Operator input and output are deleted when the customer deletes the transcript.
The customer views output in the Conversational Intelligence API or Transcript Viewer.
Compliance
Learn more about this label at nutrition-facts.ai
AI Nutrition Facts
Conversational Intelligence - Amazon Comprehend
- Description
- Language Operators with Amazon Comprehend in Conversational Intelligence to detect entities in your conversations such as products, quantities, or locations.
- Privacy Ladder Level
- N/A
- Feature is Optional
- Yes
- Model Type
- Predictive and Generative
- Base Model
- Amazon Comprehend
- Base Model Trained with Customer Data
- No
- Customer Data is Shared with Model Vendor
- No
- Training Data Anonymized
- N/A
- Data Deletion
- Yes
- Human in the Loop
- Yes
- Data Retention
- Until the customer deletes it
- Logging & Auditing
- Yes
- Guardrails
- Yes
- Input/Output Consistency
- Yes
- Other Resources
- https://www.twilio.com/docs/conversational-intelligence/pre-built-operators
Trust Ingredients
Customer data is shared with the Model Vendor for processing only. Customer data isn't used to train the Base Model.
Language Operator input and output are deleted when the customer deletes the transcript.
The customer views output in the Conversational Intelligence API or Transcript Viewer.
Compliance
Learn more about this label at nutrition-facts.ai
AI Nutrition Facts
Speech to Text Transcriptions - Programmable Voice and Conversational Intelligence
- Description
- Generate speech to text voice transcriptions (real-time and post-call) in Programmable Voice and Conversational Intelligence.
- Privacy Ladder Level
- N/A
- Feature is Optional
- Yes
- Model Type
- Generative and Predictive - Automatic Speech Recognition
- Base Model
- Deepgram Speech-to-Text, Google Speech-to-Text, Amazon Transcribe
- Base Model Trained with Customer Data
- No
- Customer Data is Shared with Model Vendor
- No
- Training Data Anonymized
- N/A
- Data Deletion
- Yes
- Human in the Loop
- Yes
- Data Retention
- Until the customer deletes
- Logging & Auditing
- Yes
- Guardrails
- Yes
- Input/Output Consistency
- Yes
- Other Resources
- https://www.twilio.com/docs/conversational-intelligence
Trust Ingredients
Conversational Intelligence and Programmable Voice only use the default Base Model provided by the Model Vendor. The Base Model is not trained using customer data.
Conversational Intelligence and Programmable Voice only use the default Base Model provided by the Model Vendor. The Base Model is not trained using customer data.
Base Model is not trained using any customer data.
Transcriptions are deleted by the customer using the Conversational Intelligence API or when a customer account is deprovisioned.
The customer views output in the Conversational Intelligence API or Transcript Viewer.
Compliance
The customer can listen to the input (recording) and view the output (transcript).
The customer can listen to the input (recording) and view the output (transcript).
The customer is responsible for human review.
Learn more about this label at nutrition-facts.ai
AI Nutrition Facts
Conversational Intelligence - PII Redaction
- Description
- Identify, redact, and extract PII in transcripts and recordings in Conversational Intelligence.
- Privacy Ladder Level
- N/A
- Feature is Optional
- Yes
- Model Type
- Predictive
- Base Model
- Private AI
- Base Model Trained with Customer Data
- No
- Customer Data is Shared with Model Vendor
- No
- Training Data Anonymized
- N/A
- Data Deletion
- Yes
- Human in the Loop
- Yes
- Data Retention
- Until the customer deletes
- Logging & Auditing
- Yes
- Guardrails
- Yes
- Input/Output Consistency
- Yes
- Other Resources
- https://www.twilio.com/docs/conversational-intelligence
Trust Ingredients
Conversational Intelligence only uses the default Base Model provided by the Model Vendor. The Base Model isn't trained using customer data.
Twilio hosts and manages its own version of the Base Model. Customer Data isn't shared with Model Vendor, it is only shared with the Twilio-hosted Base Model.
Base Model isn't trained using any customer data.
Transcripts and recordings are deleted by the customer using the Conversational Intelligence API or when a customer account is deprovisioned.
The customer views output in the Conversational Intelligence API or Transcript Viewer.
Compliance
The customer can listen to the input (recording) or view the original unredacted version, and view the output (redacted PII).
The customer can listen to the input (recording) or view the original unredacted version, and view the output (redacted PII).
The customer is responsible for human review.
Learn more about this label at nutrition-facts.ai