Artificial Intelligence Policy

From the Vice Presidency of Science, Technology and Innovation, we promote the responsible, safe, ethical, transparent, and sustainable adoption of Artificial Intelligence (AI) at Ecopetrol.

This policy guides the design, acquisition, implementation, operation, and continuous improvement of AI solutions, with the purpose of generating sustainable value, strengthening the trust of our stakeholders, and ensuring that technology serves people, safe operations, innovation, and energy transition.

In line with leading corporate sustainability assessment practices, including ESG-oriented evaluation frameworks, this policy strengthens governance, accountability, measurable impact, evidence-based management, and continuous improvement of AI-related practices.

Our purpose with AI

Artificial Intelligence at Ecopetrol seeks to generate sustainable value for the Company, its stakeholders, and society. To this end, it strengthens operational efficiency, people’s safety, asset integrity, innovation, the energy transition, and data-driven decision-making.

AI initiatives must demonstrate contribution to long-term value creation, resilience, operational excellence, sustainability performance, and stakeholder trust, supported by clear ownership, indicators, controls, and documented evidence.

Where do we apply for AI?

1. Optimization of industrial processes: using AI to maximize asset efficiency, improve operational reliability, support the integration of renewable energies, and contribute to the efficient use of resources.
2. Improving productivity and people’s quality of life: supporting work teams in repetitive or high-volume tasks, freeing up capacity for higher-value activities, always under principles of responsible use and human oversight.
3. Increasing customer and stakeholder satisfaction: enabling digital solutions and intelligent assistants that guide, serve, and respond to needs in a timely, clear, secure, and traceable manner.
4. Commitment to life: using AI to strengthen people’s safety, Industrial and Process Safety, asset integrity, and the prevention of events across the value chain.
5. Accelerating research, innovation, and knowledge generation: processing large volumes of data to produce reliable information, accelerating applied research, and support solutions for the energy transition and sustainability.

Objectives of the responsible adoption program

Within our Responsible AI Adoption Program, we have defined the following objectives:

1. Implement AI to optimize key processes and achieve greater agility, safety, efficiency, and operational profitability.
2. Create a robust infrastructure and governance standards that ensure the quality, security, traceability, and responsible use of AI across the organization.
3. Promote an ecosystem of innovative AI technologies that keeps Ecopetrol at the forefront of the energy sector, with criteria for sustainability, risk management, and value generation.
4. Strengthening Ecopetrol’s internal talent in the ethical, safe, and responsible use of AI, developing capabilities that anticipate technological change and respond to market challenges.

Establish measurable indicators, documented evidence, and periodic review mechanisms to support external sustainability assessments, ESG reporting readiness, and continuous improvement of AI governance.

Ethical principles of AI at Ecopetrol

The following principles guide the responsible use of AI at Ecopetrol and must be applied proportionally to the level of risk, impact, and criticality of each solution.

Transparency: inform and clarify to users when, how, and for what purpose advanced analytics and/or artificial intelligence are used in digital solutions.

Fairness: prevent outcomes and decisions supported by AI from generating discrimination, exclusion, or undue bias.

Risk management: assess, treat, and monitor risks associated with analytics and/or artificial intelligence throughout their life cycle.

Privacy: appropriately manage personal data and respect people’s right to privacy.

Human role: include human review in relevant steps of analytical and/or artificial intelligence processes to mitigate negative impacts and ensure responsible decisions.

Accountability for AI systems: accountability for the outcomes produced by AI systems lies with the people and areas responsible for their design, implementation, operation, and use. To request a review, report an impact, or express disagreement with the result of an AI system developed or used by Ecopetrol, the corporate channels available to employees, contractors, customers, communities, and other stakeholders must be used.

The Vice Presidency of Science, Technology and Innovation assure the Responsible Analytics and Artificial Intelligence Adoption Program at Ecopetrol. Its role includes conceptualizing and implementing strategic advanced analytics and AI projects, managing technological capabilities, promoting governance standards, ensuring secure access to platforms, protecting data and AI systems, and strengthening responsible practices across business areas.

The program also supports assessment readiness by maintaining documented governance structures, management systems, performance evidence, risk controls, and improvement actions that enable comparability, auditability, and transparent reporting on AI-related sustainability practices.

Commitments for responsible, safe, and sustainable AI

Governance and accountability: Ecopetrol has guidelines, roles, and responsibilities for the adoption of AI responsible. The Vice Presidency of Science, Technology and Innovation lead the assurance of the program, articulates technological capabilities, and promotes that user areas identify business owners, technical owners, and data owners for each AI solution. Decisions, controls, and evidence associated with AI systems must be documented throughout their life cycle.

Assessment evidence and reporting readiness: Ecopetrol shall maintain traceable documentation of policies, governance bodies, roles, risk assessments, approvals, monitoring results, performance indicators, training records, supplier controls, incidents, corrective actions, and lessons learned related to AI systems, enabling evidence-based evaluation and continuous improvement.

Life-cycle risk management: before designing, acquiring, training, deploying, or modifying an AI solution, risks related to privacy, information security, cybersecurity, bias, explainability, data quality, operational continuity, people’s rights, environmental impacts, and potential reputational or regulatory effects must be identified and managed. This management must be maintained during operation, monitoring, updating, and retirement of the solution.

Transparency and identification of AI use: when a digital solution uses advanced analytics or artificial intelligence, Ecopetrol will clearly, understandably, and proportionally inform when AI is used, for what purpose, what type of data is involved, and what channels are available to request guidance, review, or challenge results. Where applicable, AI-generated or AI-assisted content must be identified to avoid confusion, misuse, or loss of trust.

Labeling of AI-generated content and outcomes of AI-driven decisions: Ecopetrol shall visibly, clearly, and understandably identify content that has been generated, edited, or assisted by artificial intelligence when its use may influence people’s understanding, trust, or decision-making. Likewise, when a decision, recommendation, classification, prioritization, or relevant outcome is supported by AI, the existence of AI intervention must be disclosed, the purpose of the system must be explained proportionally, available channels to request human review or clarification must be indicated, and the traceability required for audit, continuous improvement, and accountability must be preserved.

Equity, non-discrimination, and bias control: AI solutions must be designed, tested, and monitored to prevent discriminatory, unfair, or biased outcomes. When bias risks are identified, mitigation, validation, and follow-up actions must be applied, especially in uses that may affect people, communities, customers, workers, suppliers, or stakeholders.

Privacy, data protection, and responsible use of information: all processing of personal data or sensitive information in AI solutions must comply with applicable regulations, corporate privacy policies, proper permission management, data minimization, authorized purposes, and the controls required to protect the confidentiality, integrity, and availability of information.

Cybersecurity and resilience: AI solutions must apply security controls by design, including access management, protection of models, data and platforms, vulnerability monitoring, prevention of unauthorized uses, incident response, and operational continuity. These controls must be aligned with corporate cybersecurity and information security guidelines.

Human oversight and responsible decisions: people remain accountable for decisions supported by AI. In high-impact or sensitive cases, human review, escalation criteria, decision traceability, and mechanisms to correct inadequate outcomes must be ensured. AI must support human judgment, not replace people’s accountability or corporate controls.

Impact measurement and sustainability: Ecopetrol will promote the impact of AI initiatives on energy efficiency, emissions reduction, responsible use of resources, operational safety, productivity, service quality, innovation, and other applicable sustainability objectives. Where feasible, benefits, risks, and lessons learned must be documented to strengthen continuous improvement and accountability.

Performance measurement and external comparability: AI initiatives should define baselines, targets, owners, frequency of measurement, and verifiable evidence for relevant indicators, including efficiency, safety, emissions, resource use, service quality, innovation, risk reduction, digital trust, and stakeholder value, to support benchmarking and sustainability performance assessments.

Talent, culture, and responsible use: Ecopetrol will strengthen the capabilities of its workers and allied teams through training, usage guidelines, good practices, and clear communication on the benefits, limits, and risks of AI. AI adoption must promote a culture of responsible innovation, continuous learning, digital security, and respect for people.

Third parties, suppliers, and acquired solutions: when Ecopetrol acquires, integrates, or uses AI solutions developed by third parties, requirements related to security, privacy, intellectual property, traceability, explainability, sustainability, support, continuity, regulatory compliance, and contractual accountability must be assessed. Likewise, suppliers must ensure compliance with the corporate standards defined for the responsible use of AI.

Supplier due diligence and accountability: third-party AI solutions must be assessed through documented due diligence covering governance, ethical use, privacy, cybersecurity, explainability, intellectual property, environmental considerations, service continuity, contractual responsibilities, audit rights, and mechanisms for remediation or escalation.

With this policy, Ecopetrol reaffirms its commitment to transforming its operations through AI in a responsible, fair, innovative, and sustainable manner, contributing to value generation, stakeholder trust, business resilience, and continuous improvement.

Related documents

Please review our privacy policy.

Please review our cybersecurity annexes.