Carbon ratings under ESMA regulation

Part 1: Questions and answers

1. What does the Regulation cover?

It covers ESG ratings issued by ESG rating providers operating in the Union (Article 2(1)). An ESG rating is an opinion or a score, or both, about a rated item’s environmental, social or governance characteristics, risks or impacts, based on an established methodology and a defined ranking system (Article 3(1)). What counts is the substance, not the label: a rating is in scope whether it is called a rating, an opinion or a score.

2. Does it apply to carbon project ratings?

CarbonAI treats its carbon project ratings as within scope of the Regulation. Each rating is based on a documented methodology and expressed on a defined rating scale. This applies whether a rating is produced by analysts, by models, or by both.

3. Does it apply only to European projects?

No. The Regulation applies to ESG ratings issued by providers operating in the Union, wherever the rated project is located.

4. What must CarbonAI do?

As part of the Regulation, CarbonAI must do the following:

• Ensure the independence of its rating activities, including from all political and economic influences or constraints (Article 15(1)).
• Adopt and implement internal due diligence policies and procedures that ensure its business interests do not impair the independence or accuracy of its ratings (Article 15(5)).
• Use rating methodologies that are rigorous, systematic, independent and capable of justification, and apply them continuously and transparently (Article 15(7)).
• Publish on its website, in a separate section, the methodologies, models and key rating assumptions behind its ratings (Article 23; see Part 2).
• Provide the required information to users of its ratings, rated items and issuers of rated items on an ongoing basis (Article 24).
• Operate under ESMA’s supervisory and enforcement powers (Articles 32 to 37).

5. What does this mean for users of CarbonAI ratings?

• Project developers: if CarbonAI rates your project, you receive the information the Regulation requires for rated items on an ongoing basis. Part 2 shows the methodology, assumptions and data sources used.
• Banks, investors, brokers and other financial firms: our published methodology and disclosures support your due diligence. If you are a regulated financial firm and refer to a CarbonAI rating in your marketing communications, the Regulation requires you to publish certain information about that rating on your website (Article 2(2)). That information is set out in Part 2.
• Buyers of carbon credits: you can check the basis of any CarbonAI rating, and our independence and due diligence obligations will be supervised by ESMA.

6. Does CarbonAI use artificial intelligence in its ratings?

Yes. Section 2.10 in Part 2 explains where AI is used in data collection and the rating process, the limitations and risks of that use, and how outputs are reviewed.

Part 2: ESG rating disclosures (Article 23)

This section sets out the methodologies, models and key rating assumptions CarbonAI uses in its ESG rating activities, as required by Article 23(1) of Regulation (EU) 2024/3005. It includes the information in point (d) of Annex I and point 1 of Annex III.

2.1 Provider and ownership

CarbonAI Data Analytics Limited, 37 Anglesea Road, Ballsbridge, Dublin D04 C4E7, Ireland. Company number 826577. LEI 9845000A67BADDC13D74. Wholly owned by CarbonAI Ltd (UK, company number 16735735).

2.2 Ratings covered

CarbonAI provides the following ratings and analytical scores through its platform:

CII, SDGII and TII are available as individual scores on CarbonAI’s dashboard and are also incorporated into the overall Carbon Project Rating methodology.

2.3 Rating objective

CarbonAI’s ratings and indices primarily assess the environmental and social impacts of carbon projects, alongside risks affecting their integrity, performance and delivery.

• Carbon Project Ratings: Assess both the positive environmental and social impacts of carbon projects and the risks associated with their integrity, implementation and delivery. These include additionality, permanence, leakage, monitoring, project status and other project-level factors.
• Country Impact Index (CII): Assesses the environmental, social and economic conditions of the host country to provide context for the potential impacts of carbon projects.
• Sustainable Development Goal Impact Index (SDGII): Measures the contribution of carbon projects to the United Nations Sustainable Development Goals, focusing on their environmental, social and economic impacts.
• Technology Impact Index (TII): Assesses the environmental and sustainability impacts associated with carbon project technologies and activities.
• Country ESG Risk Scores: Assess environmental, social and governance risks at country level, identifying potential vulnerabilities and challenges associated with each dimension. The scores provide a comparative assessment of country-level ESG risk exposure rather than measuring the potential benefits of investments.

CarbonAI’s approach combines impact assessments with project integrity and delivery risk considerations. It does not assess the financial materiality of ESG factors to investors or the financial performance of rated projects.

2.4 Scope: E, S and G

CarbonAI’s products incorporate environmental (E), social (S) and governance (G) dimensions to varying degrees, depending on their assessment objectives and methodologies.

• Carbon Project Ratings: Assess project quality, integrity and associated risks through 12 assessment features. Environmental, social and governance-related considerations are incorporated into a composite rating rather than assessed as separate ESG pillars.
• Country ESG Scores: Assess countries across environmental, social and governance dimensions, providing individual E, S and G scores alongside an aggregated ESG assessment.
• Country Impact Index (CII), Sustainable Development Goal Impact Index (SDGII) and Technology Impact Index (TII): Assess the potential benefits of investments to countries, incorporating environmental, social and economic considerations according to the relevant impact dimensions.

The scope and weighting of the underlying indicators vary by product and are defined in their respective methodologies.

2.5 Methodologies and models

CarbonAI applies structured, data-driven methodologies tailored to the objectives of each rating and index. A detailed description of the underlying indicators, scoring models, weightings, rating scales and assumptions is provided in the accompanying methodology document.

• Carbon Project Ratings: A multi-factor assessment framework combining 12 features covering project integrity, risks, sustainable development contributions and project characteristics. Individual feature scores are calculated using quantitative data, document-based assessments and AI-supported analysis. These are aggregated through a weighted scoring model to produce an overall rating from AAA (highest) to D (lowest).
• Country ESG Risk Scores: A country-level assessment framework using environmental, social and governance indicators from internationally recognised data sources. Indicators are standardised and aggregated to produce comparable country-level risk scores across the three ESG dimensions.
• Country Impact Index (CII): A country-level assessment framework using environmental, social and economic indicators to identify where investments could deliver greater relative benefits. The methodology considers country-specific conditions and development needs to calculate comparative impact scores.
• Sustainable Development Goal Impact Index (SDGII): An SDG-specific assessment framework that uses country-level development indicators to evaluate the potential relative benefits of investments targeting individual SDGs. Scores reflect country-specific needs and development gaps.
• Technology Impact Index (TII): A technology-specific assessment framework that combines country-level conditions with the characteristics and potential benefits of different technologies. Scores indicate the relative potential for a given technology to deliver positive impacts within a particular country.

Detailed methodologies, including scoring calculations, indicator selection, weighting, aggregation, rating scales and category definitions, are available in the CarbonAI Ratings and Indices Methodology Document.

2.6 Key rating assumptions

CarbonAI’s ratings and indices are based on a set of methodological assumptions that support consistent, transparent and comparable assessments.

• Data reliability: Information obtained from recognised international data providers, official registries, published project documentation and other relevant sources is assumed to be sufficiently reliable for assessment, subject to data quality checks and validation procedures.
• Data availability and completeness: Assessments are based on information available at the time of evaluation. Missing, incomplete or outdated information may affect the resulting scores and ratings and is handled according to the relevant methodology.
• Comparability: Standardised indicators, scoring criteria and normalisation procedures are used to support comparisons across projects, countries and technologies, while recognising differences in their underlying characteristics.
• Project-level assessments: Carbon Project Ratings reflect the documented characteristics, performance and risks of projects based on available evidence. Where direct verification is not possible, assessments rely on documented information and defined evaluation criteria.
• Country-level assessments: Country ESG Risk Scores and impact indices use country-level indicators as representations of broader environmental, social, economic and governance conditions. These indicators may not capture all regional or local variations.
• Potential impact: CII, SDGII and TII estimate the relative potential benefits of investments based on country-specific conditions and development needs. Higher potential impact scores do not guarantee that the anticipated benefits will be realised.
• Temporal relevance: Ratings and indices reflect the data and methodological assumptions applicable at the time of assessment. Changes in project conditions, country circumstances or newly available evidence may affect subsequent results.

The specific assumptions, their application within individual scoring models and their potential influence on rating outcomes are described in the CarbonAI Ratings and Indices Methodology Document.

2.7 Data sources

CarbonAI uses a combination of publicly available and, where applicable, non-public data to support its ratings and indices. Data sources are selected based on their relevance, reliability, coverage and suitability for the respective methodologies.

• Carbon Project Ratings: Data is obtained primarily from carbon credit registries, project design documents, monitoring and verification reports, approved methodologies, certification records and other publicly available project documentation. These sources are supplemented by country-level indicators, environmental and geospatial datasets, and, where available, additional information provided by project developers or other relevant parties.
• Country ESG Risk Scores: Data is sourced from internationally recognised organisations and public statistical databases, including the World Bank, United Nations and other authoritative providers of environmental, social, economic and governance indicators.
• Country Impact Index (CII), Sustainable Development Goal Impact Index (SDGII) and Technology Impact Index (TII): These indices primarily use country-level indicators from international organisations, official statistical sources and relevant environmental, socioeconomic and technology-related datasets. Data is selected according to the impact dimensions assessed by each index.

Data quality and missing information

CarbonAI applies data processing and validation procedures to identify inconsistencies, missing values and outdated information. Where data is unavailable, the treatment depends on the relevant methodology and may include the use of alternative indicators, defined estimation procedures or the exclusion of insufficiently supported information from the assessment.

For Carbon Project Ratings, missing or insufficient project evidence is considered in accordance with the applicable assessment criteria. The absence of information is not treated as evidence of positive project performance.

The specific data sources, data quality procedures, missing-data treatment and any estimation methods are described in the CarbonAI Ratings and Indices Methodology Document.

2.8 Industry classification

CarbonAI classifies carbon projects according to their project type, sector and applicable carbon crediting methodology, using classifications established by recognised carbon credit registries and standards. These classifications support the application of relevant assessment criteria and comparisons across projects with similar characteristics.

Country ESG Risk Scores and the Country Impact Index (CII) are country-level assessments and do not rely on industry classifications. The Sustainable Development Goal Impact Index (SDGII) is organised around the United Nations Sustainable Development Goals, while the Technology Impact Index (TII) uses technology-specific categories to assess potential benefits across countries.

Further details of the classification frameworks and their application are provided in the CarbonAI Ratings and Indices Methodology Document.

2.9 Scientific basis

CarbonAI’s methodologies are informed by established scientific research, internationally recognised assessment frameworks and empirical data. They combine quantitative analysis, statistical modelling and evidence-based evaluation approaches relevant to carbon markets, sustainable development and environmental and social risk assessment.

• Carbon Project Ratings: The assessment framework draws on established principles of carbon accounting, environmental integrity and carbon credit quality, including additionality, permanence, leakage and monitoring. Assessment criteria are informed by recognised carbon crediting standards, published research and relevant industry frameworks.
• Country ESG Risk Scores: The methodology uses internationally recognised environmental, social and governance indicators to assess country-level risks. Indicator selection and assessment approaches are informed by established sustainability and risk assessment frameworks.
• Country Impact Index (CII): The methodology draws on socioeconomic and environmental research to evaluate how country-specific conditions and development needs influence the potential benefits of investments.
• Sustainable Development Goal Impact Index (SDGII): The methodology is grounded in the United Nations Sustainable Development Goals framework and associated development indicators, using country-level evidence to assess where investments could deliver greater relative benefits.
• Technology Impact Index (TII): The methodology draws on scientific and technical evidence concerning the environmental, social and economic implications of different technologies, combined with country-specific conditions to assess their potential benefits.

CarbonAI’s methodologies are developed and reviewed by its data science team, incorporating relevant scientific literature, established assessment practices and expert input. Methodological updates are informed by new evidence, data availability and ongoing evaluation of the assessment frameworks.

Further details, including supporting scientific references, indicator selection criteria and methodological assumptions, are provided in the CarbonAI Ratings and Indices Methodology Document.

2.10 Use of artificial intelligence

CarbonAI uses artificial intelligence (AI), including natural language processing and large language models (LLMs), to support data collection, information extraction and assessment processes. AI enables the systematic analysis of large volumes of structured and unstructured information while supporting the scalability and consistency of the rating process.

Application of AI

• Data collection and extraction: AI is used to identify, extract and structure relevant information from project documentation, including project design documents, monitoring and verification reports, methodologies and other supporting materials.
• Rating assessments: AI supports the evaluation of selected qualitative features within the Carbon Project Ratings methodology. Extracted evidence is assessed against predefined criteria, while quantitative indicators and feature scores are combined using the established scoring framework.
• Information retrieval: Retrieval-Augmented Generation (RAG) is used to retrieve relevant supporting evidence from indexed documents, providing context for AI-assisted assessments and enabling traceability to underlying sources.
• Country-level ratings and indices: Country ESG Risk Scores, CII, SDGII and TII primarily rely on structured datasets and quantitative scoring methodologies rather than generative AI assessments.

Limitations and risks

AI-generated outputs may contain inaccuracies, misinterpretations, omissions or unsupported conclusions. Their reliability depends on the quality, completeness and relevance of the underlying information. Additional limitations may arise from inconsistent document formats, ambiguous terminology and variations in reporting practices.

Validation and oversight

CarbonAI applies automated data quality checks, predefined assessment criteria and review procedures to identify inconsistencies and anomalous outputs. AI-supported assessments are subject to validation and expert review where appropriate. AI does not independently determine the overall Carbon Project Rating; final ratings are generated through the established scoring and aggregation methodology.

CarbonAI periodically reviews its AI-supported processes to improve reliability, consistency and alignment with the underlying rating methodologies.

Further details of AI applications, validation procedures and methodological limitations are provided in the CarbonAI Ratings and Indices Methodology Document.

2.11 Limitations

CarbonAI’s ratings and indices are based on available data, defined methodologies and assessment assumptions. Their accuracy and completeness depend on the quality, availability and timeliness of the underlying information.

Carbon Project Ratings reflect assessments based on available project evidence and do not constitute guarantees of project performance or future carbon credit integrity. Country ESG Risk Scores and impact indices provide comparative assessments that may not capture all local conditions or future developments. In particular, impact indices assess potential benefits rather than realised outcomes.

Additional limitations relating to data availability, estimation methods and AI-supported assessments are described in Sections 2.6, 2.7 and 2.10 and in the accompanying methodology document.

2.12 Methodology changes

CarbonAI periodically reviews and updates its rating methodologies to reflect new data, scientific evidence, regulatory developments and improvements to its assessment frameworks.

The Carbon Project Ratings methodology has undergone several updates since its initial release in September 2025:

The current published methodology version, its effective date and a summary of material changes are maintained in the CarbonAI Ratings and Indices Methodology Document.

Methodological changes are reviewed to assess their potential effects on rating outcomes. Updated methodologies are published to support transparency and comparability over time.

This page reflects Regulation (EU) 2024/3005 as adopted and CarbonAI’s understanding as at 23 October 2026. It is for information only and is not legal advice. We will update it as ESMA guidance develops and whenever our methodologies change. Source: Regulation (EU) 2024/3005, EUR-Lex.