This AI-Specific Addendum ("AI Terms") is incorporated into and forms part of the VIKTOR Terms of Service and Privacy Policy. In the event of any conflict between this AI Addendum and the Terms of Service or Privacy Policy, this AI Addendum shall prevail with respect to the subject matter herein. All provisions of the Terms of Service, including those relating to changes to the service and termination, apply equally to the AI Features unless explicitly stated otherwise in this AI Addendum.
VIKTOR is committed to responsible AI deployment and helping enterprise customers adopt AI capabilities with confidence. This addendum outlines how we handle AI Features, protect your data, and support your AI governance and regulatory compliance efforts.
1.1 AI Features means any artificial intelligence or machine learning capabilities provided as part of the VIKTOR platform.
1.2 Input means any data, content, or instructions provided by you or your users to the AI Features.
1.3 Output means any result, suggestion, or content generated by the AI Features based on the Input.
1.4 Beta AI Features means any AI Features that are released in a beta, preview, early access, or similar pre-release state.
2.1 VIKTOR will handle data related to AI Features in accordance with its Privacy Policy and existing data handling practices.
2.2 By using the AI Features, you agree that VIKTOR may process your Input to:
2.3 VIKTOR may use Input to improve the AI Features by analyzing technical aspects such as usage patterns, prompt types, errors, and suboptimal agent behavior. The Input is anonymized such that it is not traceable to individual users. The analysis focuses on technical performance and functionality improvements, and does not involve substantive analysis of confidential information or proprietary content contained in the Input.
2.4 AI-related data, like all platform data, is stored and processed using cloud services provided by Amazon Web Services (AWS).
2.5 VIKTOR utilizes AWS Bedrock, a managed service that provides access to various foundation models for generative AI capabilities. AI-related data processing follows the data transfer provisions set forth in VIKTOR's Data Processing Agreement. Due to regional availability limitations of certain AI models, AI Features may involve cross-region processing within the AWS network:
All cross-region data transfers are conducted in accordance with applicable Data Protection Laws, including the use of appropriate safeguards such as Standard Contractual Clauses where required.
2.6 Data processed cross-region is encrypted in transit and at rest, processed ephemerally for the purpose of generating AI Output, and is not permanently stored outside your designated hosting region.
2.7 VIKTOR does not share your content data with external AI companies. Your data remains within our AWS environment and is used solely to provide and improve our services to you. AWS may share non-content technical usage metadata (such as API call metrics) with model providers, but your actual inputs and outputs are never shared.
2.8 For purposes of this addendum, "train" or "training" means using your Input to adjust, fine-tune, or otherwise modify the weights, parameters, or functionality of any AI model. VIKTOR does not use your Input to train AI models, nor do our AI sub-processors use your Input for training purposes. A current list of sub-processors, including those involved in AI Features, is available at www.viktor.ai/subprocessor-list.
2.9 AI-related Input and Output data is treated the same as any other customer project data stored in VIKTOR's AWS infrastructure and follows the same retention and deletion procedures. Upon termination of service or at your request, VIKTOR will delete AI-related data in accordance with the retention periods specified in your agreement. Specific deletion timelines and proof of deletion requirements can be negotiated in individual agreements.
3.1 You are solely responsible for the Input you provide to the AI Features and for your use of any Output.
3.2 You agree not to use the AI Features to process or generate any content that violates the VIKTOR Terms of Service or applicable laws.
3.3 You should review and validate any Output before using it for any purpose, as AI-generated content may be inaccurate or inappropriate for your specific use case.
3.4 AWS Bedrock employs automated abuse detection mechanisms designed to detect harmful content. If illegal content is detected, AWS may report it to appropriate authorities as required by law, as described in AWS Service Terms Section 50.12.2.
4.1 You retain all rights to your Input. By using the AI Features, you grant VIKTOR a limited license to process your Input solely to provide and improve the AI Features as described in Section 2.
4.2 You own all rights to the Output generated by the AI Features based on your Input. You grant VIKTOR a limited license to process the Output solely to provide and improve the AI Features as described in Section 2.
5.1 AI Features, including Beta AI Features, are provided "as is" without any warranties, express or implied.
5.2 VIKTOR does not guarantee the accuracy, completeness, or appropriateness of any Output for your specific needs, especially for Beta AI Features.
5.3 VIKTOR's liability related to AI Features, including Beta AI Features, is limited as set forth in the Terms of Service.
6.1 VIKTOR may offer Beta AI Features for users to test and evaluate. Beta AI Features are strictly opt-in and will not be enabled without your explicit consent. You will be informed of a feature's Beta status before opting in.
6.2 Beta AI Features are provided "as is" and may be incomplete, contain errors, or may change significantly before official release. VIKTOR does not guarantee the quality, reliability, or continued availability of Beta AI Features.
6.3 You may disable or exit Beta AI Features at any time. Upon exiting a Beta AI Feature, associated beta data will be handled in accordance with Section 2.9.
6.4 You are encouraged to provide feedback on Beta AI Features, but should not rely on these features for critical operations.
6.5 VIKTOR reserves the right to modify, suspend, or discontinue any Beta AI Features at any time. Where reasonably possible, VIKTOR will provide advance notice before discontinuing a Beta AI Feature.
6.6 Once a Beta AI Feature becomes a stable feature, it may transition to opt-out or become a standard part of the service, potentially subject to additional fees. You will be notified of such transitions in advance.
6.7 All other provisions in this AI Addendum apply to Beta AI Features unless explicitly stated otherwise in this section or in specific terms provided for a particular Beta AI Feature.
VIKTOR is committed to supporting your AI governance framework and helping you meet regulatory requirements for AI systems.
7.1 VIKTOR utilizes foundation models for which documentation is publicly available. System cards and responsible AI documentation for Claude models can be found at https://trust.anthropic.com. VIKTOR will assist in directing you to relevant documentation from AI sub-processors, including model cards, bias testing results, and other technical information needed for your compliance assessments.
7.2 VIKTOR monitors developments in AI-related regulations, including the EU AI Act, and is committed to helping you achieve compliance. We will work with you in good faith to address regulatory requirements as they become applicable to the AI Features.
This addendum is based on and consistent with the following AWS documentation and agreements:
8.1 AWS Data Processing Addendum, which governs data processing, Standard Contractual Clauses, and GDPR compliance for data transfers.
8.2 AWS Service Terms Section 50 (Machine Learning and Artificial Intelligence Services), available at https://aws.amazon.com/service-terms/, which specifies how AWS handles AI service data, including provisions for Amazon Bedrock.
8.3 AWS Bedrock FAQs (Security section), available at https://aws.amazon.com/bedrock/faqs/#security, which confirm that customer content is encrypted and stored at rest only in the customer's selected AWS region, that inputs and outputs are not shared with model providers, and that customer data is not used to train foundation models.
Last updated: March 13, 2026