Responsible AI Program

Access Control & Transparency

Lintasarta applies strict access controls across every AI capability it operates. On the Cloudeka GPU service, customer workloads are isolated through GPU virtualisation, so that one customer’s data and processes cannot be reached by another user sharing the same physical machine. The Company also enforces a Zero-Access policy through technical and administrative rules that prohibit Lintasarta personnel from accessing the contents of customer instances or the data held in customer storage.

We deliver transparency through:

• Clear labelling of AI involvement. Customer interactions with the AI Virtual Assistant are plainly identified as AI-based, so that customers and stakeholders can readily tell when information or a recommendation has been produced with AI assistance.
• Passing on model transparency documentation. For third-party AI models in our service catalogue, Lintasarta forwards the Model Card or technical fact sheet from the originating developer to the customer, so that the model’s ethical and technical limits are understood before use.
• Metadata transparency. Customers are informed that our monitoring covers operational metadata only, and never touches the content of customer AI data.
• Regular updates to users and business partners whenever there is a significant change to infrastructure, model availability in the catalogue, knowledge base scope, or data privacy policy.

Human Oversight

Lintasarta ensures that human control, monitoring, and the ability to intervene are maintained consistently throughout the AI system lifecycle. Every AI system is assigned an oversight level based on a risk assessment that weighs the impact on individuals and society, the system’s degree of autonomy, the traceability and reversibility of its decisions, and the applicable regulatory requirements. Oversight levels are matched to each service’s risk profile; for self-service offerings such as Cloudeka GPU and Cloudeka LLM, responsibility for overseeing output falls under the shared responsibility model set out in the applicable service terms.

No recommendation with material impact is left entirely to an automated system. Recommendations generated by the AI Virtual Assistant for internal purposes must be verified by authorised personnel before they are conveyed to a customer.

AI Model Integrity

To safeguard the reliability and fairness of our AI systems, Lintasarta has established the following mechanisms:

• Model distribution integrity, ensuring that every model image in the catalogue originates from an official source and has not been modified, and managing third-party model licence compliance proactively.
• Acceptance testing covering model reliability and security, detection of output anomalies and exploitation attempts, and testing of the AI Virtual Assistant’s accuracy, hallucination rate, and security.
• Fairness and non-discrimination metrics to monitor that AI systems treat all groups equitably and do not perpetuate historical bias, with corrective action taken on any disparity identified.
• An AI System Impact Assessment conducted at least once a year and repeated whenever a new model is added to the catalogue, a model version update materially changes architecture or capability, the intended purpose changes, or a new category of sensitive customer data is introduced.

Data Protection and Data Sovereignty

Lintasarta established its compliance programme for Indonesia’s Personal Data Protection Law in 2024, holds ISO 27701 certification, and has appointed a Data Protection Officer (DPO) team to steward its implementation.

In the AI context, the Company guarantees that all data processed by the LLM service and stored within customer instances remains within Indonesian jurisdiction. Customer interaction data from the AI Virtual Assistant and the content of the internal knowledge base are managed in line with the PDP Law and the Company’s information security policy, including retention limits and access restrictions.

For internal personnel, Lintasarta strictly prohibits entering Company data classified as Internal, Restricted, or Confidential into third-party generative AI tools outside the organisation’s control.

Environmental Responsibility

Lintasarta treats energy efficiency as inseparable from the provision of AI infrastructure. The Company tracks energy sustainability through Power Usage Effectiveness (PUE) measurement across its AI clusters, alongside GPU health monitoring of temperature, power draw, and physical performance to prevent both energy waste and system failure.

These initiatives form part of the Lintasarta Environmental, Social, and Governance (ESG) framework.

Grievance and Reporting Channels

We believe that customers and affected third parties have the right to question any output or decision produced by, or with the assistance of, AI. Lintasarta provides structured and readily accessible channels:

• Customer service and IT Support channels, which escalate AI-related reports to the Cyber Security Incident Response Team (CSIRT) in line with the Information Security Incident Management Procedure.
• AI Concern escalation to the Management Representative (MR), who is accountable for AIMS implementation.
• The Whistleblowing System (WBS) at www.lintasarta.net/whistleblowing-system, offering anonymous reporting and protection for reporters against retaliation.

Every report is reviewed by qualified personnel.

AI Usage and Security Training

Management provides continuing AI training that spans both technical and governance dimensions, tailored to each employee’s role. The programme covers Responsible AI Governance principles, awareness of AI risk and ethics, technical operating procedures for the Cloud GPU and LLM services, shared responsibility training, and guidance on the responsible use of the AI Virtual Assistant for customer-facing teams.

Personnel performing human oversight functions must meet defined competency requirements — including technical knowledge of the systems they oversee, the capacity to make decisions under pressure, and completion of the required training before assignment — and must maintain that competency through continuing education. Records of education, training, skills, and qualifications are retained as evidence of compliance with AIMS competency standards.

AI for Indonesian Talent Development

Giving practical effect to the principle that human and societal wellbeing takes precedence over technological advancement, Lintasarta’s commitment to responsible AI extends to building national digital talent. That commitment does not stop at our own operations. Through the AI Merdeka movement, the Company runs the Laskar AI programme, which aims to develop digital talent capable of adopting and advancing AI technology on the path to Indonesia Emas 2045 — drawing 13,588 applicants, of whom 657 were selected through a series of assessments and validation. Lintasarta also partners with the Provincial Government of West Java through the Jabar Istimewa Digital Academy (JIDA) to deliver applied training in programming, AI-based digital marketing, and networking.

Governance, Audit, and Continual Improvement

Implementation of the Lintasarta AIMS is overseen through a clear governance structure: the Board of Directors sets AI policy and objectives, the Management Representative coordinates implementation and serves as the escalation point for AI Concerns, and AI System Owners are accountable for their respective systems.

• Internal audits are conducted periodically by competent and independent personnel to assess achievement of reliability, transparency, and security objectives; the effectiveness of ISO/IEC 42001 Annex A controls; compliance with ethical standards and applicable legislation; and the validity of AI Impact Assessment results.
• Management Reviews are held at least once a year, or whenever there is a significant shift in the AI technology landscape, in regulation, or in national and global AI ethics standards.
• Continual improvement is pursued through root cause analysis of every nonconformity and AI incident, corrective and preventive action, and evaluation of the effectiveness of the actions taken.

Our objectives for AI use are aligned with human values and human rights: the wellbeing of people and society takes precedence over technological advancement, and Lintasarta AI systems are never used to manipulate, deceive, or coerce individuals.