AI Bias and Fairness Policy

This policy outlines our commitment to conducting regular bias and fairness audits on the AI models integrated into our platform, particularly those provided by third-party APIs for content creation, skill tagging, and other functionalities. Our goal is to ensure that our services are fair, ethical, transparent, and compliant with the European Union's Artificial Intelligence Act (EU AI Act), thereby fostering trust with our customers.

Scope

This policy applies to all AI-driven functionalities within our platform, including but not limited to:

Content writing assistance
Automatic content tagging with skills
Automate skills and roles taxonomy generation
Any future AI integrations

3. Compliance with the EU AI Act

3.1 Risk Assessment and Classification

Risk Categorization: We assess and classify our AI systems according to the risk categories defined in the EU AI Act:
- Unacceptable Risk: AI systems prohibited due to posing threats to safety, livelihoods, or rights of people.
- High Risk: AI systems that may significantly impact individuals' fundamental rights or safety.
- Limited Risk: AI systems requiring specific transparency obligations.
- Minimal Risk: AI systems with minimal or no risk, subject to existing legal frameworks.

3.2 High-Risk Systems Compliance

For AI systems classified as high risk, we ensure compliance with mandatory requirements:

Data Governance and Management: Implement rigorous data management practices to ensure datasets are relevant, representative, free of errors, and do not reflect unfair biases.
Technical Documentation: Maintain comprehensive technical documentation detailing the AI system's purpose, design, methodology, and testing processes.
Record-Keeping: Keep detailed logs of the AI system's activities to ensure traceability and facilitate compliance checks.
Transparency and Information Provision: Provide clear and adequate information to users about the AI system's capabilities, limitations, and intended uses.
Human Oversight: Establish effective human oversight mechanisms to monitor the AI system's operation and intervene when necessary.
Robustness, Accuracy, and Security: Ensure the AI system is robust, reliable, and secure, with appropriate measures to manage risks and prevent errors or malfunctions.

3.3 Limited Risk Systems Compliance

For AI systems classified as limited risk:

Transparency Obligations: Inform users that they are interacting with an AI system unless this is obvious from the context.
User Awareness: Provide clear disclosures when AI-generated content is presented to users.

4. Commitment to Regular Audits

Scheduled Audits: Conduct bias and fairness audits on our AI models on an annual basis, in line with the EU AI Act's monitoring requirements.
Ad-Hoc Audits: Perform additional audits in response to significant updates, identified risks, or customer feedback.

5. Audit Procedures

Data Sampling and Analysis: Analyze representative samples of AI outputs across various use cases and demographics.
Bias Detection Tools: Utilize industry-standard tools and methodologies to detect potential biases related to protected characteristics such as gender, race, age, disability, and religion.
Fairness and Non-Discrimination Assessment: Evaluate whether AI outputs comply with principles of fairness, non-discrimination, and respect for fundamental rights.

6. Documentation and Record-Keeping

Audit Reports: Document each audit in comprehensive reports detailing methodologies, findings, risk assessments, and corrective actions.
Technical Documentation: Maintain up-to-date technical documentation as required by the EU AI Act for high-risk AI systems.
Data Logs: Keep detailed records of AI system activities to ensure transparency and facilitate compliance checks.

7. Sharing Audit Findings with Customers

Transparency Commitment: In line with the EU AI Act's transparency obligations, we commit to openness regarding the fairness and compliance of our AI models.
Access to Information: Provide customers with relevant information about our AI systems, including summaries of audit reports and compliance measures, upon request.
Regular Communication: Share significant findings, updates, and improvements through official channels such as newsletters, website updates, or direct communications.

8. Mitigation Strategies

Risk Management Measures: Implement procedures to identify, assess, and mitigate risks associated with AI systems.
Corrective Actions: Develop and execute action plans promptly to address identified biases, errors, or non-compliance issues.
Collaboration with Providers: Work closely with OpenAI and other third-party providers to address issues and ensure compliance with regulatory requirements.
Continuous Improvement: Regularly update and refine our AI systems, processes, and policies to enhance fairness, reduce biases, and comply with evolving regulations.

9. User Rights and Redress Mechanisms

User Notifications: Clearly inform users when they are interacting with an AI system, especially in contexts where it may not be obvious.
Right to Contest: Provide users the ability to contest decisions made by the AI system and request human review or intervention.
Feedback Channels: Establish accessible channels for users to provide feedback, report issues, or lodge complaints regarding the AI system.
Timely Responses: Address user feedback and complaints promptly, providing appropriate remedies or explanations as required.

10. Employee Training and Awareness

Regulatory Compliance Training: Provide regular training to employees on the requirements of the EU AI Act, GDPR, and other relevant regulations.
Ethical AI Practices: Educate employees on ethical AI principles, bias detection, fairness, and non-discrimination.
Ongoing Education: Keep staff informed about updates to regulations and industry best practices.

11. Data Protection and Privacy

GDPR Compliance: Ensure all data processing activities related to AI systems comply with the General Data Protection Regulation (GDPR), including lawful data collection, processing, storage, and deletion.
Privacy by Design and Default: Incorporate data protection principles into the design and operation of AI systems from the outset.
User Consent: Obtain explicit consent from users where required and provide options for users to control their data.

12. Third-Party AI Systems and Providers

Due Diligence: Conduct thorough assessments of third-party AI providers (e.g., OpenAI) to ensure their compliance with the EU AI Act and other relevant regulations.
Contractual Agreements: Include clauses in contracts with third-party providers requiring compliance with applicable laws, regulations, and ethical standards.
Monitoring and Oversight: Monitor the performance and compliance of third-party AI systems integrated into our platform.

13. Compliance and Enforcement

Legal Compliance: Commit to full compliance with the EU AI Act, GDPR, and other applicable laws and regulations concerning AI ethics, fairness, and data protection.
Internal Enforcement: Implement internal controls and enforcement mechanisms to ensure adherence to this policy.
Penalties for Non-Compliance: Establish consequences for employees and partners who violate this policy or related regulations.

14. Contact Information

For inquiries about this policy, to request information about our AI systems, or to exercise your rights under the EU AI Act or GDPR, please contact:

Full name of legal entity: Rise To Limited trading as Learn Amp (company number 08570213)
Email address: compliance@learnamp.com
Postal address: The Old Rectory, Church Street, Weybridge, Surrey, KT13 8DE

By implementing this policy, we reaffirm our commitment to providing AI services that are fair, transparent, accountable, and compliant with the European Union's regulatory framework. We recognize the importance of upholding ethical standards and legal obligations to protect the rights and interests of our users and stakeholders.