The Principle

‍

Every major AI model on the planet has a gap it cannot see. It cannot hear the difference between Herati Dari and Kabuli Dari. It cannot distinguish Hazaragi medical terminology from standard Dari clinical vocabulary. It cannot recognize when a Pashto translation of a legal document carries a cultural connotation that reverses the intended meaning. These are not edge cases. For 6.5 million Afghan refugees and the institutions that serve them, these are the cases that matter most.

Ariana Nexus exists at the intersection where artificial intelligence meets cultural reality. We do not build AI models. We make them trustworthy. We validate their outputs, annotate their training data, moderate their content, and certify their cultural accuracy — operating within the most rigorous AI governance frameworks on earth. Every service we deliver to AI labs, technology platforms, and enterprise clients is governed by the compliance architecture described in this document.

‍

AI and Technology Services Under Compliance Governance

‍

Ariana Nexus delivers four categories of AI and technology services, each subject to the compliance frameworks described in this document:

1. AI Model Validation and Testing

Systematic evaluation of AI models — large language models (LLMs), natural language processing (NLP) systems, speech-to-text engines, machine translation systems, and multimodal AI — for accuracy, cultural appropriateness, bias, hallucination, and safety across all 24 Afghan languages and dialects.

What we test for:

• Linguistic accuracy: grammatical correctness, semantic fidelity, dialectal precision.
• Cultural appropriateness: cultural sensitivity, contextual accuracy, avoidance of culturally harmful or offensive outputs.
• Hallucination detection: identification of fabricated information, false cultural assertions, invented vocabulary, and confident-but-wrong outputs.
• Bias detection: representational bias, stereotyping, demographic skew, and systemic inaccuracy affecting Afghan-origin populations.
• Safety evaluation: outputs that could cause physical, psychological, legal, or institutional harm when used in clinical, legal, government, or educational contexts.

Compliance requirements: NIST AI RMF (Measure function), EU AI Act (conformity assessment for high-risk systems), ISO/IEC 42001 (AI Management System), Big Tech responsible AI vendor requirements.

‍

2. Human-in-the-Loop (HITL) Data Annotation and Labeling

Expert-driven data annotation, labeling, curation, and quality assurance for AI training datasets. Ariana Nexus annotators are not crowdsourced workers — they are Afghan scholars, linguists, and domain experts holding advanced degrees from institutions including Cornell, Harvard, Columbia, Oxford, and leading Afghan and European universities.

What we deliver:

• Linguistic annotation across all 24 Afghan languages.
• Cultural metadata tagging for training data contextual accuracy.
• Reinforcement Learning from Human Feedback (RLHF) with native-speaker experts.
• Quality assurance and inter-annotator agreement measurement.
• Data provenance documentation and consent chain verification.

Compliance requirements: EU AI Act Article 10 (data governance), NIST AI RMF (Map and Measure functions), ISO/IEC 42001, client-specific DPA and data handling requirements, GDPR where training data contains EU personal data.

‍

3. Content Moderation and Trust & Safety

Afghan-language content moderation for AI platforms, social media, messaging applications, and user-generated content environments. Ariana Nexus moderators identify and classify hate speech, violent extremism, misinformation, disinformation, harassment, child safety violations, and culturally specific harmful content in Dari, Pashto, Hazaragi, and other Afghan languages.

What we deliver:

• Policy-aligned content classification across client-defined trust and safety taxonomies.
• Cultural context analysis — content that appears benign in translation may carry threatening, extremist, or harmful meaning in Afghan cultural context.
• Escalation and reporting for content requiring law enforcement or regulatory referral.
• Moderator wellbeing protocols, including exposure management and psychological support.

Compliance requirements: EU Digital Services Act (DSA), EU AI Act (general-purpose AI obligations), UK Online Safety Act, NIST AI RMF (Govern function), platform-specific content policy frameworks, GDPR for EU user data.

‍

4. Cultural Intelligence API (Planned Product)

A software product under development that provides programmatic access to Ariana Nexus's cultural validation capabilities — translation validation, cultural appropriateness scoring, dialect identification, and document verification for Afghan languages. Usage-based pricing model.

Compliance requirements (planned): EU AI Act (provider obligations for AI systems), NIST AI RMF, ISO/IEC 42001, SOC 2 Type II (when API is production), API security standards (OWASP), GDPR data processing requirements.

Target: MVP scoping Q2 2026; development phase Q3–Q4 2026; beta launch 2027.

‍

U.S. AI Governance Frameworks

‍

NIST AI Risk Management Framework (AI RMF 1.0)

The NIST AI Risk Management Framework, published in January 2023, is the primary U.S. government framework for managing AI risk. Ariana Nexus aligns its AI services with all four AI RMF functions:

GOVERN — Organizational AI Risk Governance:

• Ariana Nexus maintains a documented internal AI Governance Policy that establishes roles, responsibilities, and oversight for AI-related activities.
• AI risk management is integrated into organizational leadership decision-making.
• AI ethics principles are communicated to all personnel involved in AI service delivery.
• Third-party AI engagements are governed by Engagement Agreements and Data Processing Agreements that define permitted uses, data handling requirements, and quality obligations.

MAP — AI Risk Identification and Context:

• Before any AI engagement, Ariana Nexus conducts a risk mapping assessment that evaluates the AI system's intended use context, affected populations, potential harm vectors, data sources and quality, and regulatory classification (high-risk, limited-risk, minimal-risk under EU AI Act).
• Afghan-language AI systems are assessed for the specific risks inherent to low-resource languages: limited training data, dialect conflation, cultural bias amplification, and hallucination in knowledge-sparse domains.

MEASURE — AI Risk Assessment and Evaluation:

• Ariana Nexus conducts structured evaluation of AI model outputs using expert annotators, standardized evaluation rubrics, and quantitative metrics (accuracy, F1 score, BLEU score, cultural appropriateness score, hallucination rate).
• Evaluation results are documented in AI Validation Reports delivered to clients with full methodology, metrics, findings, and recommendations.
• Bias assessments evaluate demographic representation, stereotyping, and systemic accuracy disparities.

MANAGE — AI Risk Treatment and Monitoring:

• Identified risks are communicated to clients with remediation recommendations.
• For ongoing engagements, Ariana Nexus provides periodic re-evaluation to track model performance over time, detect degradation, and identify emerging risks.
• Incident response procedures apply to AI-specific incidents, including model failure, data poisoning detection, and cultural harm events.
  ‍

White House Executive Order on AI (EO 14110, October 2023 — Revoked January 20, 2025)

‍

EO 14110 ("Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence") was revoked on January 20, 2025, and superseded by EO 14179 ("Removing Barriers to American Leadership in Artificial Intelligence," January 23, 2025). EO 14179 does not include comparable testing, red-teaming, or transparency provisions. Ariana Nexus maintains its AI safety and governance practices as institutional best practice independent of the current federal executive posture. The capabilities developed in alignment with EO 14110 continue to serve clients:

• Red-teaming and safety testing: Ariana Nexus provides cultural and linguistic red-teaming for AI systems, identifying failure modes specific to Afghan languages and cultural contexts.
• AI transparency: Ariana Nexus's AI Validation Reports provide documented evidence of model evaluation methodology and findings, supporting client compliance with transparency requirements.
• Bias and discrimination prevention: Ariana Nexus's bias detection services identify and document discriminatory outputs affecting Afghan-origin populations, supporting client compliance with the EO's civil rights protections.

NIST AI 100-2: Adversarial Machine Learning

NIST AI 100-2 (Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations) provides a framework for understanding AI security threats. Ariana Nexus's AI services address several threat categories:

• Data poisoning: Ariana Nexus's data annotation quality assurance processes detect compromised or malicious training data.
• Model evasion: Ariana Nexus's validation identifies inputs designed to cause model misclassification or failure in Afghan-language contexts.
• Privacy attacks: Ariana Nexus's data governance ensures that personal data in training datasets is handled in accordance with GDPR, HIPAA, and client-specific requirements.
  ‍

International AI Governance Frameworks

‍

European Union AI Act (Regulation (EU) 2024/1689)

The EU AI Act is the world's first comprehensive AI regulation. It establishes risk-tiered obligations for AI providers, deployers, importers, and distributors. Ariana Nexus aligns its AI services with the EU AI Act across all applicable risk tiers:

High-Risk AI Systems (Annex III):

Where Ariana Nexus provides validation, annotation, or testing for AI systems classified as high-risk under the EU AI Act — including systems used in healthcare (medical devices, clinical decision support), biometric identification, critical infrastructure, education, employment, law enforcement, immigration, and administration of justice — Ariana Nexus applies the following requirements:

• Data Governance (Article 10): Training, validation, and testing datasets are subject to quality measures including examination for bias, representativeness assessment, gap analysis, and data preparation procedures appropriate to the intended purpose.
• Technical Documentation (Article 11): Ariana Nexus provides documentation of its validation and annotation methodology, metrics, findings, and data governance practices to support client compliance with Article 11 technical documentation requirements.
• Transparency (Article 13): Ariana Nexus's AI Validation Reports support client compliance with transparency obligations by documenting model performance characteristics and limitations.
• Human Oversight (Article 14): Ariana Nexus's HITL processes provide the human oversight layer that high-risk AI systems require. All AI data processed by Ariana Nexus undergoes human review at critical quality, bias, and cultural accuracy checkpoints.
• Accuracy, Robustness, Cybersecurity (Article 15): Ariana Nexus evaluates model accuracy and robustness across Afghan-language inputs, including adversarial and edge-case testing.
  ‍

General-Purpose AI Models (Chapter V):

Where Ariana Nexus provides services to providers of general-purpose AI models (GPAIs) — including large language models deployed by AI labs and Big Tech companies — Ariana Nexus supports compliance with:

• Article 53: Technical documentation requirements for GPAI models.
• Article 55: Systemic risk obligations for GPAI models with systemic risk, including adversarial testing and red-teaming.
  ‍

UNESCO Recommendation on the Ethics of Artificial Intelligence (2021)

‍

The UNESCO Recommendation is the first global normative instrument on AI ethics, adopted by 193 member states. Ariana Nexus aligns its AI services with UNESCO's core values and principles:

• Human Rights and Human Dignity: AI services are designed to respect and promote the rights of individuals, particularly vulnerable and marginalized populations including Afghan diaspora communities.
• Living in Peaceful, Just, and Interconnected Societies: AI validation detects and mitigates outputs that could promote division, discrimination, or harm.
• Diversity and Inclusiveness: Ariana Nexus's coverage of all 24 Afghan languages — many of which are severely underrepresented in AI training data — directly supports the inclusiveness principle.
• Environment and Ecosystem Flourishing: Assessed for relevance in future sustainability-focused AI engagements.
• Proportionality and Do No Harm: AI validation evaluates whether model outputs could cause disproportionate harm to Afghan-origin populations.
• Transparency and Explainability: AI Validation Reports provide transparent documentation of evaluation methodology and findings.
• Fairness and Non-Discrimination: Bias detection and mitigation services directly support this principle.
• Human Oversight and Determination: HITL processes ensure human oversight of all AI data processing.
• Privacy and Data Protection: Data governance aligns with GDPR, HIPAA, and NIST AI RMF.
• Responsibility and Accountability: Ariana Nexus maintains documented accountability for AI service quality and compliance.
  ‍

OECD AI Principles (2019, updated 2024)

‍

The OECD AI Principles, endorsed by over 46 countries, establish five principles for trustworthy AI. Ariana Nexus aligns its services with:

• Inclusive Growth, Sustainable Development, and Well-Being: Afghan-language AI inclusion supports equitable access.
• Human-Centred Values and Fairness: Bias detection and cultural validation uphold fairness.
• Transparency and Explainability: Documented validation methodology and reporting.
• Robustness, Security, and Safety: Model testing for accuracy, adversarial robustness, and safety.
• Accountability: Documented governance, auditable processes, and client reporting.
  ‍

Council of Europe Framework Convention on AI (2024)

‍

The first legally binding international AI treaty. Ariana Nexus monitors its ratification and implementation timeline and aligns its AI governance practices with the Convention's requirements for human rights protection, democratic values, and rule of law in AI systems.

‍

G7 Hiroshima AI Process Code of Conduct (2023)

‍

The G7 voluntary code of conduct for organizations developing advanced AI systems. Ariana Nexus's AI validation and red-teaming services support client compliance with the code's commitments to pre-deployment safety testing, post-deployment monitoring, and transparency.

‍

Big Tech and AI Lab Compliance Requirements

‍

Vendor Qualification Standards

AI labs and Big Tech companies maintain rigorous vendor qualification processes for AI data services providers. Ariana Nexus is designed to meet or exceed these requirements:

Data Security — Typical expectation: SOC 2 Type II, ISO 27001, encryption at rest and in transit. Ariana Nexus: SOC 2 roadmap (2027); ISO 27001 roadmap (2027); AES-256 encryption operational.

Data Handling — Typical expectation: DPA execution, data retention/deletion procedures, audit rights. Ariana Nexus: DPA template ready; Purview DLP and Sensitivity Labels operational; audit logging active.

Annotator Qualification — Typical expectation: Demonstrated subject-matter expertise, linguistic proficiency. Ariana Nexus: Advanced-degree holders from Cornell, Harvard, Oxford, and leading institutions.

Quality Assurance — Typical expectation: Inter-annotator agreement metrics, QA sampling, error analysis. Ariana Nexus: Structured QA process with documented IAA measurement.

Bias and Fairness — Typical expectation: Bias assessment methodology, demographic representation analysis. Ariana Nexus: Bias detection framework with Afghan cultural context expertise.

Confidentiality — Typical expectation: NDA execution, information security controls. Ariana Nexus: NDA required for all personnel; Purview IRM and DLP enforced.

Compliance — Typical expectation: GDPR, CCPA, sector-specific regulations. Ariana Nexus: Privacy Policy covers GDPR, CCPA, HIPAA; four-tier data classification.

Responsible AI — Typical expectation: Alignment with client's responsible AI principles. Ariana Nexus: Internal AI Governance Policy; NIST AI RMF alignment.

Worker Wellbeing — Typical expectation: Moderator wellbeing programs, exposure management. Ariana Nexus: Content moderator wellbeing protocols planned.

Scalability — Typical expectation: Ability to scale annotation workforce. Ariana Nexus: Human Intelligence Collective scaling plan (50+ interpreters, 10 PMs).

‍

Platform-Specific Considerations

Ariana Nexus is positioned to serve AI platforms and technology companies across the ecosystem:

Frontier AI Labs (Anthropic, OpenAI, Google DeepMind, Cohere, ElevenLabs): Cultural and linguistic validation of LLM outputs, RLHF annotation with native-speaker experts, red-teaming for Afghan-language failure modes, training data quality assurance, and safety evaluation for deployment in healthcare, government, and legal contexts.

Big Tech Platforms (Microsoft, Google, Amazon, Meta, Apple): Content moderation for Afghan-language trust and safety, AI model localization testing, cultural appropriateness validation for product features, and linguistic QA for voice assistants, translation services, and search algorithms.

Enterprise AI Adopters: AI governance consulting, model validation for healthcare NLP, government AI system compliance assessment, and cultural bias auditing for enterprise AI deployments.

‍

Internal AI Governance

‍

Ariana Nexus maintains a documented internal AI Governance Policy that governs the organization's own use of AI tools and its delivery of AI services to clients:

Governance Structure

• AI Governance Authority: The CEO and Founder holds ultimate authority over AI governance decisions, with support from domain leads for healthcare, government, and technology engagements.
• AI Ethics Review: Material AI service engagements — particularly those involving high-risk AI systems, vulnerable populations, or government applications — undergo an ethics review before engagement acceptance.
• Policy Scope: The AI Governance Policy applies to all Ariana Nexus personnel, contractors, and Collective members who participate in AI service delivery.
  ‍

Core Principles

The Ariana Nexus AI Governance Policy is built on six principles:

1. Human Primacy: AI augments human judgment. It does not replace it. Every AI data pipeline operated by Ariana Nexus incorporates human-in-the-loop oversight.

2. Cultural Integrity: AI systems that serve Afghan populations must reflect Afghan cultural, linguistic, and ethical realities. Ariana Nexus does not validate AI models that systematically misrepresent, erase, or harm Afghan cultural identity.

3. Transparency: Ariana Nexus provides transparent documentation of its AI service methodology, metrics, and findings to every client. No black-box processes.

4. Accountability: Every AI service engagement has a named engagement lead who is accountable for quality, compliance, and ethics. Deliverables are traceable and auditable.

5. Safety First: If an AI system poses a credible risk of harm — to patients, to legal proceedings, to national security, or to vulnerable populations — Ariana Nexus reports the risk to the client and recommends remediation before continued deployment.

6. Data Stewardship: Client data is processed only for the purposes specified in the Engagement Agreement. No repurposing. No unauthorized retention. No use of client data to train Ariana Nexus's own models without explicit written authorization.
‍

Prohibited AI Practices

Ariana Nexus does not participate in or support:

• Development or deployment of AI for mass surveillance of populations.
• AI systems designed for social scoring or profiling of individuals based on ethnicity, religion, or national origin.
• Autonomous weapons systems or lethal autonomous AI.
• AI-generated content intended to deceive, manipulate, or defraud.
• AI training on data obtained through exploitation, coercion, or violation of consent.
  ‍

AI Data Governance

Data Provenance and Chain of Custody

Every dataset processed through the AI Data Factory is subject to documented provenance tracking:

• Source identification: Origin of the data, including whether it was client-provided, publicly sourced, or internally generated.
• Consent chain verification: Where the data contains personal data, the consent basis (or other lawful basis) for processing is documented.
• Annotation history: Every annotation, label, quality assessment, and modification is logged with the annotator identity, timestamp, and version.
• Classification: All AI data is classified under the four-tier data classification framework (Public, Internal, Confidential, Restricted). AI data containing PII is classified as Restricted.
  ‍

Data Quality Standards

• Representativeness assessment: Training datasets are assessed for demographic, dialectal, and contextual representativeness to minimize bias.
• Inter-annotator agreement (IAA): Multiple annotators evaluate the same samples to measure consistency. IAA metrics are reported to clients.
• Error analysis: Systematic analysis of annotation errors to identify patterns, training needs, and process improvements.
• Continuous monitoring: For ongoing engagements, data quality is monitored over time to detect drift, degradation, or emerging bias.
  ‍

Client Data Isolation

• Client AI data is stored in dedicated SharePoint environments with engagement-specific Security Groups.
• Data from different clients is never commingled.
• Annotators are assigned to specific engagements and do not have access to data from other engagements.
• Client data is not used for Ariana Nexus's internal purposes without explicit written authorization.
  ‍

Alignment with AI and Technology Compliance Frameworks

‍

Ariana Nexus's AI and technology compliance architecture is designed in alignment with the following recognized frameworks and standards:

NIST AI RMF 1.0 — Govern, Map, Measure, Manage functions for AI risk. Aligned — all four functions implemented in AI service delivery.

NIST AI 100-2 — Adversarial machine learning threats and mitigations. Aligned — data poisoning, evasion, and privacy threat awareness.

White House EO 14110 (revoked January 20, 2025) — Safe, secure, and trustworthy AI development. Historical alignment maintained — red-teaming, safety testing, bias prevention, and transparency practices retained as institutional best practice; superseded by EO 14179 (January 23, 2025).

EU AI Act (Regulation (EU) 2024/1689) — Risk-tiered AI regulation. Aligned — high-risk, GPAI, and transparency requirements addressed.

EU Digital Services Act — Content moderation obligations for online platforms. Aligned — content moderation services support client DSA compliance.

UK AI Regulation (Pro-Innovation Framework) — Principles-based AI regulation. Aligned — safety, transparency, fairness, accountability principles.

UNESCO AI Ethics Recommendation (2021) — Global AI ethics norms (193 member states). Aligned — all 10 principles addressed.

OECD AI Principles (2019, updated 2024) — International AI policy framework (46+ countries). Aligned — all 5 principles addressed.

Council of Europe AI Convention (2024) — First binding international AI treaty. Monitoring — ratification and implementation tracking.

G7 Hiroshima AI Process (2023) — Voluntary code of conduct for advanced AI. Aligned — validation and red-teaming support code compliance.

ISO/IEC 42001:2023 — AI Management System standard. Roadmap (2028) — certification planned.

ISO/IEC 23894:2023 — AI Risk Management guidance. Aligned — risk management follows ISO 23894 structure.

ISO/IEC 5338:2023 — AI System Lifecycle Processes. Aligned — lifecycle considerations applied to service delivery.

HIPAA (for healthcare AI) — PHI protection in AI training and validation. Compliant — as Business Associate under executed BAAs.

GDPR (Article 22, Recital 71) — Automated decision-making and profiling. Aligned — human oversight ensured in all AI processes.

CCPA/CPRA — Consumer rights regarding automated decision-making. Aligned — data subject rights honored for AI data.

‍

What AI Compliance Means for Our Clients and Partners

‍

For AI labs (Anthropic, OpenAI, Google DeepMind, Cohere, ElevenLabs): Ariana Nexus provides compliance-ready cultural and linguistic validation, RLHF with expert annotators (not crowdsourced), red-teaming for Afghan-language failure modes, and documented AI Validation Reports that support your responsible AI disclosures, EU AI Act compliance, and safety testing requirements.

For Big Tech platforms (Microsoft, Google, Amazon, Meta, Apple): Our content moderation and trust & safety services cover Afghan languages with cultural context expertise that automated systems cannot replicate. Our annotator qualification standards, data governance, and security controls are designed to meet your vendor qualification requirements.

For enterprise AI adopters: Our AI validation services assess your deployed AI systems against NIST AI RMF, EU AI Act, and industry-specific regulations (HIPAA for healthcare AI, NIST 800-171 for government AI). We provide documented findings with remediation recommendations.

For government agencies deploying AI: Our services support AI safety testing, NIST AI RMF adoption, and bias prevention — practices we maintain as institutional best practice regardless of federal executive order status. Note: EO 14110 was revoked January 20, 2025; Ariana Nexus maintains these capabilities because they serve client interests and align with international AI governance frameworks (EU AI Act, NIST AI RMF, OECD AI Principles). Our Afghan-language expertise addresses the specific AI equity concerns relevant to government programs serving refugee and immigrant populations.

If your organization requires AI compliance documentation, vendor qualification materials, or an AI governance briefing, contact trust@ariananexus.com or +1 (202) 771-0224.

‍

Maturity Roadmap

‍

Ariana Nexus views AI and technology compliance as a multi-year journey. The following roadmap reflects our planned maturation path:

Phase 1: Foundation (Current — 2026) — Operational

Internal AI Governance Policy documented. NIST AI RMF alignment across all four functions. AI model validation and HITL annotation services operational. Content moderation capability for Afghan languages. Data provenance and quality assurance processes. Four-tier data classification for AI data. Client data isolation and DPA framework.

Phase 2: Hardening (Q3–Q4 2026) — In Planning

Cultural Intelligence API MVP scoping and development. AI Validation Report template standardization. Moderator wellbeing program formalization. Big Tech vendor qualification documentation package. EU AI Act compliance assessment toolkit for clients.

Phase 3: Product Launch (2027) — Planned

Cultural Intelligence API beta launch. SOC 2 Type II audit (AI service controls). Formal AI red-teaming methodology publication. Big Tech and AI lab engagement portfolio.

Phase 4: Certification (2028) — Planned

ISO/IEC 42001 AI Management System certification. EU AI Act conformity assessment participation. Cultural Intelligence API production launch. AI Data Platform (curated Afghan linguistic datasets) scoping.

Phase 5: Advanced (2029–2030) — Planned

Automated cultural validation engine. Real-time AI bias monitoring dashboard. Multi-language expansion beyond Afghan languages. Integration with client AI governance platforms.

Phase 6: Long-Horizon (2030+) — Vision

AI-native cultural intelligence infrastructure. Autonomous AI oversight systems with human escalation. Global AI compliance automation across 100+ jurisdictions. Quantum-AI security evaluation.

‍

Limitation of Liability and Disclaimers

‍

AI Services, Not AI Systems. Ariana Nexus provides AI validation, annotation, content moderation, and compliance services. Ariana Nexus does not develop, deploy, or operate AI models or AI systems. Clients retain full responsibility for the development, deployment, use, and governance of their own AI systems. Ariana Nexus's services support client AI governance but do not substitute for independent AI risk management.

No Guarantee of AI Model Performance. Ariana Nexus does not warrant or guarantee the accuracy, safety, fairness, or fitness for any particular purpose of any AI model validated, tested, or trained using data processed by Ariana Nexus. AI model performance depends on factors beyond Ariana Nexus's control, including model architecture, training methodology, deployment environment, and use context.

Framework Alignment vs. Certification. Where this page states "aligned," controls are designed in accordance with the framework but formal certification has not been obtained unless stated otherwise.

Roadmap Items. The maturity roadmap reflects current plans as of the Effective Date. Roadmap items are forward-looking statements, not binding commitments. Ariana Nexus reserves the right to modify roadmap items at its sole discretion.

Client-Specific Obligations. Specific AI service obligations are defined in the applicable Engagement Agreement and Data Processing Agreement. In the event of conflict between this page and an Engagement Agreement, the Engagement Agreement controls.

Limitation of Liability. TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, ARIANA NEXUS'S TOTAL AGGREGATE LIABILITY FOR ALL CLAIMS ARISING OUT OF OR RELATED TO AI AND TECHNOLOGY SERVICES SHALL NOT EXCEED THE AMOUNTS SET FORTH IN THE APPLICABLE ENGAGEMENT AGREEMENT, OR, WHERE NO ENGAGEMENT AGREEMENT EXISTS, ONE HUNDRED DOLLARS ($100). ARIANA NEXUS SHALL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, PUNITIVE, OR EXEMPLARY DAMAGES ARISING FROM OR RELATED TO AI SERVICES, MODEL VALIDATION, DATA ANNOTATION, CONTENT MODERATION, OR AI GOVERNANCE SUPPORT. NOTHING IN THIS SECTION SHALL LIMIT OR EXCLUDE ARIANA NEXUS'S LIABILITY FOR: (A) FRAUD OR FRAUDULENT MISREPRESENTATION; (B) DEATH OR PERSONAL INJURY CAUSED BY NEGLIGENCE; OR (C) ANY OTHER LIABILITY THAT CANNOT BE EXCLUDED OR LIMITED BY APPLICABLE LAW, INCLUDING BUT NOT LIMITED TO LIABILITY UNDER THE UK UNFAIR CONTRACT TERMS ACT 1977, THE UK CONSUMER RIGHTS ACT 2015, OR GDPR.

Dispute Resolution. Any dispute arising out of or relating to this page shall be subject to the dispute resolution provisions in the Terms of Use, Section 18.

‍

This page is provided for informational purposes and does not constitute a warranty, guarantee, or binding commitment regarding Ariana Nexus's AI governance or compliance posture. Capabilities described herein are subject to change. Nothing in this page shall be construed as a waiver of any right, defense, or immunity available to Ariana Nexus under applicable law.

‍