Responsible Artificial Intelligence Policy

I. Purpose

To regulate the application of artificial intelligence (“AI”) technology across the business scenarios of Concord New Energy Group Limited (the “Group”), to promote the compliant, safe, prudent, transparent and sustainable development of AI technology, and to prevent risks arising from the application of such technology, the Group hereby formulates this Policy. This Policy is intended to establish the principal code of conduct to be observed by the Group in its application of AI technology.

II. Application

This Policy applies to business activities in which the Group currently or may in the future build, procure, integrate, configure, use or manage AI systems, tools and related services, including but not limited to internal management, office collaboration, knowledge-based Q&A, data analysis and identity verification, as well as other business scenarios where AI technology is applied following assessment.

III. Commitments

The Group commits to strictly complying with all laws and regulations relating to AI in the jurisdictions where it operates, carrying out AI management and continuously improving its AI management system.

Data Privacy Protection

The Group respects and protects the privacy of personal data involved in the use of AI technology, strictly complies with the laws and regulations on information security and privacy protection in the jurisdictions where it operates, and implements necessary management and protection measures in respect of the relevant data.

Cybersecurity and Information Security Protection

The Group brings AI systems, models, tools and related services within the scope of its information security and privacy protection management system. It adopts security mechanisms such as encryption of core data and least-privilege access, establishes an information security incident handling mechanism, and regularly conducts vulnerability scanning and training to strengthen emergency response capabilities, thereby guarding against security threats including malicious attacks, data tampering and model theft.

Avoiding Algorithmic Bias

In the design, development, procurement, integration, configuration or use of AI technology, the Group identifies and mitigates potential bias and prevents the generation of unfair bias or discrimination. For AI applications involving individual rights and interests, customer rights and interests, or significant business judgments, the Group regularly assesses fairness, and conducts prudent evaluation and ensures proper use. It promptly detects deviations and takes necessary corrective measures to ensure that AI decisions and outputs meet fairness requirements.

Human Oversight and Intervention

For matters involving major business decisions, financial disclosure, compliance judgments, legal affairs, personal safety and other highly sensitive areas, the Group upholds the principle of final human decision-making. Control measures are applied according to the scenario and risk level of the matter; all key decisions must undergo human review before being implemented, and the Group retains the right of human intervention over AI throughout the process.

Transparency and Explainability

The Group’s AI applications follow the principles of transparent disclosure, understandability and traceability. Information on the use of AI systems, their capability boundaries, limitations, data sources and associated potential risks is disclosed to users, enabling stakeholders to understand how AI systems affect their decisions or outcomes. Where AI is used in the course of interacting with users, appropriate means shall be used to inform them that they are interacting with an AI system rather than a human being. For outputs and decisions generated by AI, explanations matched to the needs of stakeholders are provided.

Responsibility and Accountability

The Group has established an accountability management structure for AI applications, specifying the personnel responsible for AI projects and the scope of their duties, including risk assessment, compliance review and the implementation of corrective measures, so as to ensure that AI application processes are traceable, manageable and accountable. An investigation and handling mechanism has been established for errors or harm caused by AI, ensuring that responsibility can be traced to the specific party concerned.

Boundaries of AI Capability

The Group clearly defines the authorized scope of use, capability boundaries and limitations of AI according to different business scenarios, and formulates the necessary policies and safeguards to prevent misuse, task deviation or unintended application.

Environmental Sustainability

The Group advocates taking ecological and environmental impacts into account in the application of AI. When selecting AI infrastructure and service providers, and subject to meeting requirements on security, compliance, quality and business needs, the Group incorporates energy efficiency indicators and the proportion of green energy used into its assessment criteria, giving priority to solutions featuring high energy efficiency, low resource consumption and minimal environmental impact. Where relevant scenarios are involved, the Group requires its own or third-party AI data centers and model training environments to achieve reasonable levels of efficiency in energy consumption, water use and carbon emissions.

Prohibited Uses

The Group prohibits the use, deployment or development of the following AI systems or applications:

1. those employing subliminal, deliberately manipulative or deceptive techniques that materially distort the behaviour of individuals;
2. those exploiting vulnerabilities relating to age, disability or a specific socio-economic situation that materially distort behaviour;
3. those used for social scoring or social ranking, or that apply differential treatment based on inappropriate criteria;
4. those conducting biometric identification, facial recognition surveillance or other biometric monitoring activities without lawful authorization;
5. other AI systems and applications that violate laws and regulations, regulatory requirements, ethical norms and the Group’s internal rules.

IV. Management Measures

For sensitive AI capabilities such as facial recognition and surveillance-based identification, the Group conducts assessments based on applicable laws and regulations, information security requirements and business necessity, and strictly limits the purpose of use, applicable scenarios and scope of authorization. Access to the back-end of the relevant systems and to their data is granted only to authorized personnel, and the granting, modification and revocation of such access shall correspond to job responsibilities and actual management needs. Access controls shall be applied to the facial data, identification records and other sensitive information involved, which shall not be exported, shared or used for any purpose beyond the authorized scope without approval.

The Group applies distinct labelling to AI-generated content or AI-assisted outputs according to the actual scenario, and, for interactive AI systems used by internal users, discloses their artificial intelligence nature at appropriate points in the interaction. Outputs of internal AI-assisted tools serve only as an office reference. Where matters involve the application of internal rules, business judgments or management decisions, the Group’s official documents, departmental rules and regulations, and the conclusions of human review shall constitute the final basis, and AI-generated content shall not be used as the sole basis for decision-making.

The Group has put in place mechanisms to detect and correct drift or degradation of AI models over time. It conducts baseline benchmarking and performance tracking of the output accuracy and knowledge currency of its internal AI-assisted tools, using user feedback and reports of anomalous results to identify signals of model drift or degradation. Where any decline in performance or staleness of knowledge is detected, the Group promptly updates its knowledge base, refreshes retrieval sources, or retrains/recalibrates the model, so as to prevent model capability from eroding over time. Deployed AI applications involving automated decision-making, or that may affect the rights and interests of individuals, customers, or third parties, are subject to regular assessments for fairness and bias to ensure their fairness and freedom from discrimination.

The Group monitors the ecological footprint of AI data centers and models. Having regard to the future scale of AI application and procurement, together with suppliers’ service capabilities, green computing capacity and energy consumption management as well as the Group’s own sustainable development requirements, the Group will continuously track the energy consumption and environmental impacts that may arise from the use of AI tools.

The Group has established a feedback and appeal communication process in respect of AI use. For AI-assisted tools used internally, where outputs are inconsistent with formal policies, documents issued by the Group or established business facts, users may report the matter through internal communication channels to the department responsible for the policy concerned, the information management department or the system administrator. Following human verification, the knowledge sources will be corrected, system configurations adjusted, or the matter referred to the supplier for resolution.

The Group recognises the enabling value of AI tools for sustainable development. When expanding the scope of AI application in the future, and in conjunction with sustainability-related scenarios such as production and operations optimization, energy consumption management and data analysis, the Group will progressively develop a quantitative indicator system covering dimensions such as scope of use, working hours saved, reduction in repeated queries and rate of improvement in management efficiency, so as to assess the actual contribution of AI tools to operational efficiency gains and to the achievement of sustainability targets, and to better understand the role of AI technology in delivering environmental and social benefits.

The Group conducts training on the compliant use of AI tools, covering risks that may arise in the course of AI application, including privacy leakage, improper uploading of sensitive or confidential information and questionable accuracy of generated content. Employees are expressly required to exercise prudence in the use of AI tools and are strictly prohibited from uploading or disclosing internal sensitive materials, trade secrets and personal information to AI tools that have not passed compliance verification. AI-generated content must be subject to human verification and shall not be used as the sole basis for the interpretation of policies, business decisions or external information disclosure.

V. Policy Review

The Group will conduct necessary reviews of the Policy at regular intervals. The ESG Committee oversees and supervises the implementation of this Policy and maintains timely updates to it.