Version 20260525

This AI Terms Addendum (“AI Addendum”) is incorporated into and forms part of the Main Services Agreement (“MSA”) between Uniphore Technologies North America, Inc. (“Uniphore”) and Customer. This AI Addendum sets forth terms specific to Uniphore’s AI Features. Customer’s use of the AI Features is also subject to the Acceptable Use Policy. In the event of any conflict between this AI Addendum and the Agreement, this AI Addendum will control solely with respect to AI-specific subject matter not otherwise addressed in the Agreement, subject to the order of precedence set forth in the Agreement. Capitalized terms not defined herein have the meanings set forth in the Agreement.

1. DEFINITIONS

“AI Features” means the artificial intelligence, language model, natural language processing, and other AI-powered capabilities embedded in or made available through the Services, including but not limited to transcription, summarization, sentiment analysis, agent assistance, agent verification, agentic workflow creation, model fine-tuning, and conversational AI features. The AI Features are part of the Services.

“AI Model” means any large language model or small language model deployed by Uniphore as part of the Services that can generate Outputs.

“Fine-Tuned Model” means an AI Model customized using Customer Data or Customer-provided data.

“Hallucination” means any Output that is factually incorrect, fabricated, misleading, or does not accurately reflect the input data or real-world facts.

2. AI MODELS AND SERVICE MODIFICATIONS

2.1 Third-Party Models and Service Updates. Some AI Features are powered by pre-trained models provided by third-party AI model providers. Uniphore selects, manages, and may replace these providers, as well as its own AI Models and AI Features, at any time to maintain and improve the Services. Uniphore will provide reasonable advance notice of any change that it reasonably believes could materially and adversely affect the functionality of the Services as described in the Documentation. Customer will comply with any applicable usage restrictions imposed by third-party AI model providers, as described in the Documentation. Uniphore is not responsible for the inherent capabilities, limitations, or outputs of any third-party pre-trained models used within the Services.

2.2 Model Performance and Drift. AI Model performance may degrade over time as real-world data distributions shift. Uniphore monitors its AI Models for material performance degradation and will notify Customer if such degradation materially impacts the Services.

2.3 Transparency Information. Upon Customer’s written request, Uniphore will make available, to the extent reasonably practicable: (a) a description of the categories of training data used to develop Uniphore’s proprietary AI Models; and (b) known material limitations and performance characteristics of AI Features as documented in the Documentation. Uniphore is not obligated to disclose information that would reveal trade secrets, proprietary model architecture, or confidential third-party information. Any information so disclosed is Uniphore’s Confidential Information.

3. DATA AND TRAINING

3.1 No Training on Customer Data. Uniphore will not use Customer Data to train, fine-tune, or improve AI Models that are made available to other customers, consistent with the Agreement.

3.2 Fine-Tuning. Customer may fine-tune AI Models using the Services, and Uniphore may use Customer Data to develop Fine-Tuned Models at Customer’s direction. Uniphore does not provide Customer with the underlying model weights or architecture of any pre-trained model or Uniphore-proprietary model component, unless otherwise expressly agreed in writing. Uniphore has no liability for the performance or quality of any customer-provided pre-trained model, Fine-Tuned Model, SLM, or the outcomes of any automated fine-tuning cycles that Customer has configured or approved.

4. OUTPUTS

4.1 Output Accuracy Disclaimer. Outputs are inherently probabilistic and subject to factual errors and hallucinations, inaccuracies, and biases characteristic of current AI technology. Customer is solely responsible for reviewing, validating, and deciding whether to rely on any Output. OUTPUTS ARE NOT AND WILL NOT BE CONSTRUED AS LEGAL, MEDICAL, FINANCIAL, OR OTHER PROFESSIONAL ADVICE.

4.2 Human Review. Customer will implement and maintain appropriate human review of Outputs, including: (a) review by a qualified human before any Output is used to make a decision that affects an individual’s rights, opportunities, or welfare; (b) regular auditing of Output quality for Customer’s specific use case; and (c) mechanisms for correcting or overriding erroneous Outputs. Human review responsibility rests solely with Customer, and Uniphore has no liability for Customer’s failure to implement it.

4.3 Confidence Scores and Accuracy Characteristics. Where AI Features surface confidence scores or accuracy indicators, such scores are probabilistic estimates only, may be inaccurate, and do not guarantee Output accuracy. Accuracy may vary based on input quality, domain specificity, and other factors specific to Customer’s environment and use case.

4.4 Similarity of Outputs. Due to the nature of AI systems, the Services may generate similar or identical Outputs for different customers. Customer will not assert any claim against Uniphore or another customer based on Output similarity.

4.5 Third-Party IP in Outputs. Outputs may in some instances resemble third-party content. Customer is responsible for reviewing Outputs for potential third-party intellectual property concerns before use, particularly before incorporating Outputs into Customer-facing or publicly distributed materials.

5. AI GOVERNANCE AND REGULATORY COMPLIANCE

5.1 Compliance Responsibility. Customer is responsible for ensuring that its use of the Services and any Customer Artifacts comply with all applicable laws, regulations, and industry standards, including the EU AI Act and applicable sector-specific regulations. Where the EU AI Act or equivalent legislation applies, Customer is solely responsible for all deployer obligations thereunder. Customer will not use the Services for any purpose prohibited by applicable AI regulation, including subliminal manipulation, exploitation of vulnerabilities, real-time remote biometric identification in publicly accessible spaces where prohibited, and social scoring by public authorities. Uniphore will cooperate in good faith to provide information about the Services reasonably necessary for Customer’s compliance assessments. Any information so disclosed is Uniphore’s Confidential Information.

5.2 Regulatory Change. If a change in applicable law, regulation, or binding regulatory guidance materially affects Uniphore’s provision of the AI Features, Uniphore may modify the Services or this AI Addendum as reasonably necessary to comply. The parties will cooperate in good faith to address any material impacts to Customer’s use of the Services.

6. CUSTOMER ARTIFACTS AND DEPLOYMENT

Customer bears sole responsibility for all aspects of the development, testing, deployment, and operation of Customer Artifacts, including SLMs, Fine-Tuned Models, and AI Agents. Uniphore has no liability for the performance, quality, or operation of Customer Artifacts.