AI Transparency & Governance

1. Purpose and Scope

1.1 Purpose

This addendum provides comprehensive information regarding MVPR's use of artificial intelligence technologies in service delivery, including AI model specifications, governance frameworks, transparency commitments, and compliance measures. It expands upon Section 7 of the SLA regarding AI-enabled services.

1.2 Integration with Other Documents

The addendum should be read alongside:

• Section 7 of the SLA (AI-Enabled Services)
• Section 8 of the MSA (Data Protection and Customer Data)
• The Data Processing Agreement
• MVPR's Operations Security Policy
• MVPR's Secure Development Policy

1.3 Updates and Versioning

MVPR conducts quarterly reviews and updates reflecting material changes in AI capabilities, models, or regulatory requirements. Updated versions are distributed via email notifications and published at this page.

2. AI System Architecture and Models

2.1 AI Models Currently Deployed

2.2 Model Selection Criteria

MVPR evaluates AI models based on:

1. Task suitability - matching model capabilities to specific use cases
2. Quality and accuracy - demonstrated performance on relevant benchmarks
3. Security and privacy - provider's data handling and security practices
4. Compliance - provider's adherence to GDPR and EU AI Act
5. Cost-effectiveness - balancing performance with operational efficiency
6. EU data residency - preference for EEA-based processing

2.3 Model Update and Change Management

Notification timelines for material changes:

• Major model migrations: 30 days advance notice
• Version updates: notification with monthly service updates
• Security patches: immediate implementation with retroactive notification

2.4 Model Training and Data Usage

Customer data is not used to train AI models without explicit written consent. Key commitments:

• Zero-data retention by Anthropic and OpenAI for model training purposes
• No model fine-tuning on customer data unless explicitly contracted
• Customer retains all intellectual property rights to inputs and outputs

3. AI-Enabled Workflow and Human Oversight

3.1 Content Creation Workflow

MVPR employs a structured human-in-the-loop workflow with seven distinct stages, each with a mandatory human checkpoint:

3.4 When AI Is Not Used

MVPR does not use AI for:

• Strategic planning and decision-making
• Client relationship management
• Legal or compliance advice
• Final publishing decisions
• Contract or business negotiations
• Crisis communications requiring human expertise

4. Quality Control and Accuracy

4.1 Quality Assurance Framework

4.2 Accuracy and Fact-Checking

Core principles: zero tolerance for fabrication; all factual claims verified against authoritative sources; immediate correction protocol for identified errors.

4.3 Known AI Limitations and Mitigation

AI models may generate plausible but incorrect information, lack real-time knowledge due to training cutoffs, misinterpret nuanced contexts, and exhibit biases present in training data. MVPR mitigates these through human editorial review, supplementary research for recent events, and contextual briefings for nuanced situations.

5. Bias Detection and Fairness

5.2 Bias Mitigation Strategies

• Editorial teams trained to identify and correct biased outputs
• Cultural sensitivity review for diverse audience content
• Prompts designed to elicit balanced perspectives and avoid stereotypes
• Regular review of AI provider commitments to fairness and diverse training data

5.3 Fairness in Journalist Targeting

Our journalist database and targeting algorithms:

• Prioritise relevance based on beat coverage and past work
• Avoid discriminatory filtering by protected characteristics
• Provide transparency in matching logic

Prohibited: filtering journalists by gender, race, ethnicity, or any protected characteristic; algorithmic redlining or systematic exclusion of media outlets.

5.4 Reporting and Remediation

Bias concerns can be reported directly to your Account Director. We investigate within 3 business days, correct the specific output, identify systemic issues, and update prompts or workflows to prevent recurrence. MVPR conducts annual internal bias audits; results available on request.

6. Explainability and Transparency

6.2 Customer's Right to Explanation

Upon request, MVPR provides:

• Identification of specific AI models used for particular deliverables
• Explanation of AI's contribution to specific outputs
• Human review and oversight activities performed
• Alternative approaches considered
• Limitations or uncertainties in AI-generated components

Submit requests to your Account Director or tom@mvpr.io. We respond within 5 business days.

7. Data Privacy and AI Processing

7.1 Data Flow in AI Systems

Primary storage is in the EEA (Google Cloud Platform EU multi-region). AI processing occurs via EEA entities. No systematic transfers to non-EEA jurisdictions.

7.2 Data Minimisation in AI Processing

• Only data necessary for the specific AI task is submitted
• Personal identifiers are anonymised where not required
• Irrelevant data is stripped from prompts
• AI providers do not retain customer data post-processing

7.3 AI and Sensitive Data

MVPR does not process Sensitive Personal Data (racial/ethnic origin, political opinions, health data, biometric data under GDPR Article 9) through AI models unless explicitly authorised in writing by the customer.

8. Intellectual Property and AI-Generated Content

8.1 Ownership of AI Outputs

The customer is the exclusive owner of all AI-generated content produced under this agreement, including:

• All AI-generated drafts, outlines, and suggestions
• All human-refined versions of AI outputs
• All final deliverables incorporating AI components
• All modifications, updates, and derivatives

MVPR retains no proprietary rights to customer's AI-generated content.

8.2 Third-Party IP Considerations

All AI-generated content is run through plagiarism detection tools and reviewed by editors trained to identify derivative works. Anthropic and OpenAI provide IP indemnification for AI-generated outputs covering claims that outputs infringe third-party intellectual property rights (subject to provider terms).

9. Compliance and Regulatory Alignment

9.1 EU AI Act Compliance

MVPR's AI use falls primarily into low-risk or minimal-risk categories under the EU Artificial Intelligence Act. Our systems do not involve critical infrastructure, education, employment decisions, law enforcement, or biometric identification.

9.2 GDPR Compliance

Key article compliance:

• Article 22: MVPR's AI does not make solely automated decisions with legal or significant effects - all decisions involve human oversight
• Article 25: Privacy-by-design principles embedded in Secure Development Policy
• Article 32: ISO 27001:2022 certified
• Article 28: Formal Data Processing Agreement with Standard Contractual Clauses

10. Limitations and Customer Responsibilities

10.1 Known Limitations of AI Systems

AI systems may:

• Generate plausible but factually incorrect information (hallucinations)
• Lack real-time information due to knowledge cutoffs
• Misinterpret complex or nuanced contexts
• Exhibit biases present in training data
• Produce varying outputs for similar inputs
• Produce formulaic outputs without human refinement

10.2 Customer's Role and Responsibilities

Customers should:

• Carefully review all AI-assisted deliverables before approval or publication
• Supply detailed briefings, brand guidelines, and contextual information
• Ensure use of AI-generated content complies with applicable laws
• Accept ultimate responsibility for published content

10.3 Prohibited Uses

Customers agree not to:

• Use services to generate content violating applicable laws or regulations
• Use services to generate harmful, misleading, or inappropriate content
• Attempt to reverse-engineer MVPR's AI implementation
• Use AI-generated content to train competing AI models or services

11. Incident Response and Issue Escalation

11.2 Response Timelines by Severity

12. Governance and Accountability

12.1 AI Governance Structure

12.3 AI Ethics Principles

1. Human-centric approach - AI augments human expertise; humans retain decision-making authority
2. Transparency - clear disclosure of AI's role and capabilities
3. Fairness - proactive bias detection and mitigation
4. Privacy - data minimisation and privacy-by-design
5. Accountability - clear lines of responsibility for AI outcomes
6. Safety - robust quality control and risk management
7. Sustainability - responsible use of AI resources

13. Customer Support and Resources

Contact points for AI-related support:

• General inquiries: tom@mvpr.io
• Account support: your Account Director
• Security/privacy concerns: escalate through your Account Director

14. Amendment and Version Control

MVPR reviews this addendum quarterly. For material changes, we provide 30 days advance notice via email. Continued service use after the effective date constitutes acceptance.

15. Contact Information