BN Digital – AI Services Summary
Scope & Nature of Services
- AI systems are probabilistic; outputs may vary, be inaccurate, incomplete, biased, or unexpected.
- Applies to AI consulting, LLM integration, AI agent development, generative AI, MLOps, and AI-enabled automation.
- Supplements service-specific packages and the Quality, Retention, Security & Compliance document.
Included by Default
- AI/ML architecture design and technical guidance.
- Integration of third-party AI models and APIs (e.g. OpenAI, Anthropic, Google, open-source models) using official documentation.
- Development of AI agents, pipelines, and automation workflows.
- Prompt engineering and model behaviour configuration for defined use cases.
- Test-environment validation of AI outputs against agreed acceptance criteria.
- Standard data pipeline development for AI input/output.
- Code-level integration of AI components into new or existing systems.
Excluded Unless Agreed in Writing
- Model training, fine-tuning, and dataset curation.
- Advanced performance benchmarking beyond basic functional checks.
- Bias, fairness, or explainability assessments.
- Regulatory compliance assessments (e.g. EU AI Act, GDPR, sector rules).
- Post-handover monitoring, retraining, or drift management.
- Content moderation or safety evaluation frameworks.
- Legal review of AI-generated outputs.
- Ongoing human review/oversight of AI content.
Client Responsibilities
- Clearly define use case, target users, and operational context.
- Review and validate AI outputs before any production, commercial, or public use.
- Implement human review for consequential decisions.
- Ensure compliance with all applicable AI and data regulations.
- Obtain necessary consents for data used as AI inputs.
- Manage and secure third-party AI API keys and credentials.
- Comply with third-party AI provider terms of service.
BN Digital is not responsible for harm from client deployments lacking adequate review, validation, or oversight.
AI Output Disclaimer
- No warranty that AI outputs are accurate, complete, appropriate, unbiased, or hallucination-free.
- Outputs may be factually wrong, contextually inappropriate, or legally non-compliant.
- Client bears full responsibility for reviewing, validating, and owning AI-generated content, decisions, and actions.
BN Digital disclaims liability for:
- Use of AI outputs in medical, legal, financial, safety-critical, or regulated contexts without professional review.
- Harm from reliance on AI outputs without human validation.
- Unintended, offensive, or harmful AI-generated content.
- Behaviour in edge cases or adversarial conditions outside agreed tests.
Intellectual Property & Data
Deliverables
- Upon full payment, client receives rights to original code, configurations, and architecture created for them, subject to pre-existing IP and third-party components (per Development Service Package).
AI Model Outputs
- BN Digital does not claim ownership of AI-generated content.
- No warranty that outputs are free from third-party IP claims or copyright issues.
- Client is responsible for IP due diligence before commercial use.
Training Data
- For training, fine-tuning, or RAG using client data, client warrants it has all necessary rights and that use is lawful.
- Client indemnifies BN Digital against claims arising from use of client-provided training data.
Third-Party AI Providers
- Frequent reliance on providers like OpenAI, Anthropic, Google DeepMind, Mistral, Hugging Face.
- BN Digital is not responsible for:
- Uptime, availability, or performance of third-party APIs.
- Changes to model behaviour, capabilities, or pricing.
- Policy changes affecting the client’s use case.
- Third-party data handling practices.
- Client must review and accept provider terms and data processing agreements.
- Post-handover remediation due to provider changes is out of default scope.
Regulatory Compliance
- AI is subject to evolving laws (EU AI Act, GDPR, sector rules, national laws).
- BN Digital does not provide legal or regulatory advice.
- Client is solely responsible for compliance and should seek independent legal counsel, especially in regulated sectors.
Data Privacy & AI
- For personal data processing, client must ensure compliance with relevant data protection laws (e.g. GDPR, CCPA).
- Where required, client must:
- Obtain and document user consent for AI-assisted processing.
- Apply data minimisation and purpose limitation.
- Honour data subject rights (e.g. explanation, erasure) in AI-enabled systems.
- Execute a Data Processing Agreement with BN Digital if BN Digital acts as a processor.
Ethical Use
BN Digital may decline or stop engagements involving:
- Unlawful surveillance, profiling, or social scoring.
- Automated decisions with material impact on individuals without human oversight.
- Disinformation, deepfakes, or manipulative content.
- Weapons, harmful automation, or legally prohibited applications.
Client represents that their intended AI use does not fall into these prohibited categories.
Acceptance
- Deliverables are accepted when:
- Client approves in writing, or
- Five (5) business days pass after delivery with no documented objection.
- Acceptance confirms functional alignment with scope, not ongoing accuracy of AI outputs.
Limitation of Liability
- BN Digital’s total liability is capped at the total fees paid for the specific engagement.
- No liability for indirect, incidental, consequential, special, or punitive damages (e.g. reputational harm, fines, third-party claims, business losses).
- Client assumes all liability for decisions, actions, or communications based on AI outputs.
- Quality, Retention, Security & Compliance
- Acceptance & Handover Protocol
- Scope Change & Change Request Policy
- Development Service Package
BN Digital – AI Services Summary
BN Digital AI Services – Summary
Nature of AI Services
- AI/ML systems are probabilistic and non-deterministic.
- Outputs may be inaccurate, incomplete, biased, or unexpected.
- Client accepts this as a condition of engagement.
Included by Default
- AI/ML architecture design and technical guidance.
- Integration of third-party AI models and APIs (per official docs).
- Development of AI agents, pipelines, and automation workflows.
- Prompt engineering and model configuration for defined use cases.
- Test-environment validation against agreed acceptance criteria.
- Standard data pipelines for AI input/output handling.
- Code-level integration of AI components into software systems.
Excluded Unless Agreed in Writing
- Model training, fine-tuning, dataset curation.
- Advanced performance benchmarking.
- Bias, fairness, or explainability assessments.
- Regulatory compliance assessments (e.g., EU AI Act, GDPR).
- Post-handover monitoring, retraining, or drift management.
- Content moderation or safety evaluation frameworks.
- Legal review of AI outputs.
- Ongoing human review/oversight of AI content.
Client Responsibilities
- Define use case, users, and operational context.
- Review and validate AI outputs before production/public use.
- Implement human review for consequential decisions.
- Ensure compliance with all applicable AI laws and regulations.
- Obtain necessary consents for data used with AI.
- Manage and secure third-party AI API keys and credentials.
- Comply with third-party AI providers’ terms of service.
BN Digital is not responsible for harm from client deployment without adequate review or oversight.
