Data Use and Model Training Addendum
Effective Date: 1 July 2026
This Data Use and Model Training Addendum explains how TurboHelp may use Customer Data, De-identified Data, Anonymized Data, Aggregated Data, and Usage Data to operate, improve, evaluate, and train TurboHelp services and models.
1. Relationship to Other Terms
This addendum supplements the TurboHelp Terms of Service, AI Product Terms, Data Processing Addendum, Product Privacy Notice, and Privacy Policy.
If this addendum conflicts with the Data Processing Addendum regarding personal data processing, the Data Processing Addendum controls for that processing.
2. Customer Data Ownership
Customer owns raw Customer Data. TurboHelp does not claim ownership of raw Customer Data.
Customer Data includes support conversations, chat logs, messages, tickets, help center content, documents, prompts, AI Inputs, AI Outputs, user feedback, handoffs, workflow traces, resolution status, escalation reasons, and human corrections.
3. License to Operate and Improve
Customer grants TurboHelp the right to access, process, transmit, store, analyze, display, and use Customer Data to provide, secure, support, troubleshoot, operate, analyze, and improve the Services.
This includes use of Customer Data to:
- Generate, route, summarize, classify, and evaluate support interactions.
- Retrieve and rank relevant knowledge sources.
- Configure and improve Turbo's support behavior.
- Troubleshoot product issues and support Customer.
- Detect abuse, fraud, spam, security threats, and operational problems.
- Measure performance, reliability, latency, quality, and usage.
- Develop and improve product features, workflows, analytics, and safety controls.
Customer is responsible for having all rights, notices, consents, permissions, and lawful bases needed to provide Customer Data to TurboHelp and permit this processing.
4. De-identified, Anonymized, Aggregated, and Usage Data
TurboHelp may create De-identified Data, Anonymized Data, Aggregated Data, Usage Data, and derived analytics from Customer Data.
TurboHelp may use this data during and after the customer relationship to operate, analyze, improve, train, fine-tune, evaluate, benchmark, secure, and support the Services and models.
TurboHelp owns De-identified Data, Anonymized Data, Aggregated Data, Usage Data, and derived analytics that do not identify Customer, Permitted Users, End Users, or any individual.
TurboHelp will use reasonable measures designed to avoid re-identifying De-identified Data or Anonymized Data, except where necessary to test safeguards, investigate abuse, comply with law, or validate data protection controls.
5. Default Model Training Authorization
Unless Customer opts out as described below, Customer authorizes TurboHelp to use Customer Data and data derived from Customer Data to train, fine-tune, evaluate, benchmark, and improve models and AI systems used to provide or improve the Services. This includes Customer Data about End Users, support conversations, chat logs, messages, tickets, help center content, prompts, Input, Output, feedback, handoffs, workflow traces, resolution status, escalation reasons, and human corrections.
Training and evaluation may include:
- Retrieval and ranking models.
- Classification models.
- Summarization models.
- Escalation and handoff models.
- Feedback parsing and quality models.
- Safety filters and abuse detection.
- Product analytics and performance benchmarks.
- AI response evaluation and regression testing.
TurboHelp may prioritize De-identified Data, Anonymized Data, Aggregated Data, and Usage Data for training and evaluation, but some model-improvement workflows may use Customer Data unless Customer opts out.
Customer acknowledges that resolved support interactions, human corrections, feedback, and escalation outcomes are especially useful for improving support automation quality. Unless Customer has opted out, TurboHelp may use those signals to improve TurboHelp models for Customer and for other TurboHelp customers, provided TurboHelp does not disclose raw Customer Data to other customers.
6. Training Opt-Out
Customer may opt out of use of Customer Data for model training by emailing [email protected] or using an in-product opt-out control if available.
TurboHelp will apply the opt-out within a commercially reasonable period after receiving the request and verifying the Account. The opt-out applies prospectively unless TurboHelp expressly agrees otherwise in writing.
The opt-out does not require TurboHelp to delete or unwind models, weights, evaluations, analytics, or derived learnings created before the opt-out became effective, provided they do not identify Customer, Permitted Users, End Users, or any individual.
The opt-out does not limit TurboHelp's use of:
- Usage Data.
- Aggregated Data.
- Security logs and abuse-prevention data.
- Billing and account data.
- Diagnostic and performance data.
- Operational analytics needed to provide, secure, support, analyze, and improve the Services.
- De-identified Data or Anonymized Data where use is permitted by law and contract.
7. Third-Party LLM Provider Training
TurboHelp may use third-party LLM providers to generate Output or support AI functionality.
TurboHelp will contractually restrict third-party LLM providers from using Customer Data to train or improve their own models. Third-party LLM providers may process Input and Output only to provide, secure, monitor, or support AI functionality, subject to applicable subprocessor terms.
TurboHelp's right to use Customer Data for TurboHelp's own model training does not permit third-party LLM providers to train or improve their own models on Customer Data.
8. Customer-Specific Configuration
TurboHelp may use Customer Data to personalize or configure Turbo for Customer's Account, including knowledge retrieval, escalation behavior, instructions, workflow rules, and response style.
Customer-specific configuration is intended to improve Customer's own use of the Services and may continue while the Account remains active or as otherwise described in the applicable retention settings.
9. Feedback and Human Corrections
User feedback, agent corrections, resolved conversations, escalation reasons, quality ratings, and workflow traces are Customer Data. TurboHelp may use them to evaluate performance, identify gaps, improve retrieval, test quality, and improve AI behavior as described in this addendum.
Customer should not include sensitive regulated data in feedback or corrections unless TurboHelp has expressly agreed in writing.
10. Data Minimization and Safeguards
TurboHelp will use commercially reasonable safeguards for model-improvement workflows. Safeguards may include access controls, logging, data filtering, de-identification or anonymization techniques, retention controls, security review, and evaluation procedures.
No de-identification or anonymization method is perfect. TurboHelp will use reasonable measures appropriate to the nature of the data and the intended use.
11. No Sensitive Regulated Data
Customer must not submit Protected Health Information, medical records, diagnosis or treatment data, health insurance data, payment card data, government IDs, children's data, or other sensitive regulated data unless TurboHelp expressly agrees in writing. See the No-PHI / No-HIPAA Policy.
12. Customer Notices
Customer is responsible for updating its privacy policy, AI disclosures, terms, and End User notices to describe its use of TurboHelp, AI-assisted support, conversation processing, cookies or tracking, and service improvement.
The Product Privacy Notice includes a sample End User disclosure snippet Customer may adapt.