| 1 Introduction | 7 How to submit a response
8 Evaluation Criteria and Procedure Annex 2 – Mandatory text in responses |
1 Introduction
MPAI is planning on developing a standard that enables a human or an AI System to know about the parameters that characterise an AI System vis-à-vis its users and the operational context in which the AI is applied. This information should be provided in a standard format that is expected to be called “AI Passport”.
This Call for Technologies includes some of the Functional Requirements that the AI Passport should support and seeks to acquire the technologies that will enable MPAI to develop the planned Technical Specification: AI Passport.
2 Definitions
Table 1 – Terms and definitions
| Term | Definition |
| Agency | An entity appointed by an authority to assign Identifiers to AI Systems. |
| AI Passport | A dataset conveying information related to a specific AI System. |
| AI system | A machine-based system that, for explicit or implicit objectives, infers from the inputs it receives how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. AI systems may operate with different levels of autonomy and adaptiveness (OECD). |
| Certification | The formal attestation, issued by an authority, that an AI system conform to a defined set of requirements (technical, legal, ethical, safety, and risk-management standards). |
| Fingerprint | A function characterised by the properties of bit sensitivity, and anti-collusion or uniqueness that applies to a file containing a trained AI with its parameters. |
| Ground Model | A pre-trained model or a model architecture that may be used as a basis for training. |
3 Problem statement
Currently, regulators are developing legally or administratively binding regulations and some standardisation organisations are developing procedural rules related to AI Systems and their use. Examples are Approved Regional regulations such as the EU AI Act and ISO/IEC 4200x AI Governance.
The released deliverables, however, do not offer solutions to the use cases described in the following section.
Currently, regulations lack practical applicable procedures and standards lack specifications for practical operation, leaving this to the initiative of individual users – companies/entities.
The functional requirements identified in section 5 are intended to provide practical response to the use cases mentioned in the following.
4 Use Cases
4.1 Human-AI interaction
- A user is searching for an AI System with appropriate functionalities. By looking at the AI Passport, the user can easily find the main features of a particular AI System.
- A user can be guaranteed that an AI satisfies a set of of requirements (e.g., technical, legal, ethical, safety).
- A user wants to know the ground model and the datasets used for training a given AI System.
- A user is assured that a given AI System is the one that they intend to use.
- A user wants to use an AI System of which it knows the functionality but is unsure that the AI used with specific data respects the regulations of their country.
- A user wants to know the KPIs (generic and application-specific) of an AI System.
- A user wants to know the properties of the input and output data and the protocols used by the AI System simply by looking at the AI Passport.
- A developer wants to state the main characteristics of an AI System in a syntactic and semantically bounded context.
- Companies developing a contract related to an AI System want to make reference to an undisputed source of technical and legal information.
- Regulatory bodies want to enforce a common way of expressing information about AI Systems.
4.2 Machine-to-AI interaction
A machines wants to know the features of an AI System before dealing with it:
- Search for an AI System with specific characteristics, e.g., identification of a suitable AI System in a marketplace/platform/local repository. This may be performed in non-real or real-time.
- Direct AI-to-AI interaction. An AI System can decide whether to enter in a relationship with another AI-System on the basis of the latter’s characteristics, e.g., available input and output protocols, computational requirements, compliance with specific regulations.
5 Functional Requirements
5.1 General requirements
Table 2 includes two columns, the first provides the entries of the AI Passport and the second the functional requirements.
Table 2 – Functional Requirements for AI Passport
| AI Passport ID | – ID of an AI Passport associated to a specific AI Instance
– ID shall be provided by – an authorised agency – the AI developer if an agency. – An ID identifies an AI model with bit-wise precision to the entries of the AI Passport. – An AI may have more than one ID. |
| Agency ID | ID of the Passport-issuing entity (an agency or a developer if it is an agency). (Taxonomy) |
| AI System ID | A string that uniquely identifies and AI System. |
| Function | Natural Language Text describing what the AI is intended to do. |
| Fingerprint | Hash used to validate that the AI instance is what it says it is. It includes a specification (e.g., an ID) of the fingerprint method. Note that if the AI changes, the hash is different. |
| Interfaces | The interfaces of the AI System |
| – Input/Output | May be Input, Output, or Input/Output (Taxonomy) |
| – Protocol(s) | The Protocols used to input and output data. (Taxonomy) |
| Data | Data types used by the AI System |
| – Semantics | A standard way to represent a relevant set of characteristics of a data instance (syntax and semantics) (Taxonomy or identification system or annotation). For instance, MPAI Data Type web specifications include the Semantics of the data. |
| – Governance | The legislation and regulations under which the data is input and data was produced. |
| Certification | The legislation, regulations, or standards under which the operation of the AI was certified (Taxonomy) |
| Origin | Information on AI creation |
| – Ground Model | – Shall be identified according to a standard identification system for AI Models. The AI obtained by training a Ground Model shall include its own ID and the Ground Model’s (Parent) ID.
– Based on Agencies whose role is to provide identifiers to the models. |
| – Training | – The name and version of the data set (as set by the data set developer)
– A reference (e.g., DOI, URL, hash), DAO (Decentralized Autonomous Organisation) – Reference to developer documentation |
| KPIs | – |
| – General | – List of application-independent KPIs for the AI System, e.g., Precision, Area Under the Curve. |
| – Application | – Lists of application-independent KPIs for the AI.
– Guidelines for app-dependent KPIs may be developed. |
| Ethical Evaluation | Evaluations of the Ethical aspects of the AI System. |
| – Accountability | All data reference the meaning of the Accountability Taxonomy. |
| – Environmental | All data reference the meaning of the Environment Taxonomy. |
| – Social (Dual use) | All data reference the meaning of the Social Taxonomy. |
| Technology Environment | List of environment elements affecting the AI System. |
| – Hardware Architecture | Shall reference a Taxonomy (Vendor, Hardware type)
All data reference the Hardware Architecture Taxonomy. |
| – Run-time environment | Shall reference a Taxonomy (Developer, Name, Version, OS)
All data reference the Run-time environment Taxonomy. |
5.2 Notes on AI-Passport binding with the AI System
Possible methods
- Method 1
- Create a macro-object that includes the AI and the Passport.
- Watermark the AI with a payload that contains the Passport ID.
- Method 2
- Use of a DAO System to store the AI Passport.
- Method
- Use of a Blockchain-based system to ensure a truly distributed governance.
Proposal for AI-Passport binding are requested.
6 List of requests by the Call
6.1 Taxonomies
The Taxonomies requested by this call are:
- Agencies
- Can be public or privately owned
- Is appointed by a relevant authority
- Provides access to a data base of registered AI Systems.
- Records in the data base shall provide the data of the requested AI Passport
- Access to a Passport may be subject to restrictions based on jurisdiction
- Legislations (for certification, data governance)
- Regulations (for certification, data governance)
- Standard bodies
- Standards
- Input/output Data
- Input/output protocols
- KPI (general, application). Each KPI has
- KPI name,
- Target application
- Reference specification defining KPI.
- Methods that take (accountability, environmental and social) issues into account
- Name
- Reference (e.g., DOI, URL, official doc),
- How it is adopted by the AI (by design, bias, and ex-post KPI).
- Hardware architecture (by vendor, hardware type)
- Run-time environment (Developer, Name, Version, OS)
A Proposal for a Taxonomy should include
- The procedure to identify entities in a Taxonomy
- The Taxonomy template (the Taxonomy should be extensible).
- The proposed Taxonomy with actual entries, if one is possible at the time a response is provided.
- The procedure for extending the Taxonomy
6.2 Other requests
Standard identification system for AI Models.
Methods for binding AI Passports to AI Systems.
Formats to represent AI Passport.
7 How to submit a response
Those planning to respond to this Call are:
- Informed that the AI Passport Call will be presented at two online events held on 2026/9/10 at 08:00 UTC and 18:00 UTC. The recorded events and related material will be published at [7].
- Requested to communicate their intention to respond to the MPAI secretariat by 2026/09/27 with an initial filled out version of the form in Annex 1. Submissions of Annex 1 help MPAI to properly plan for reviewing the responses. However, an Annex 1 submission is not a commitment to submit a Response and a Response without a previous submission of Annex 1 will still be accepted.
- Advised to visit the MPAI-AIP web page regularly where relevant updates may be posted.
- Strongly encouraged to use the Template for Responses [6] when submitting a response to this Call.
- Submit a Response by 2026/10/27T15:00 UTC.
A respondent is requested to present their Response at an online meeting properly announced by the MPAI Secretariat to the respondent. If the presenter of a submission does not attend that meeting, their Response will be discarded.
Respondents are advised that, upon acceptance by MPAI of their submission in whole or in part for further evaluation, MPAI will require that a non-MPAI member immediately join MPAI. If that Respondent elects not to join, their submission will be discarded. Directions on how to join MPAI are available here.
Further information on MPAI can be obtained from the MPAI website.
8 Evaluation Criteria and Procedure
The following process will be used to assess the Responses:
- An Evaluation Panel is created from:
- MPAI members in attendance.
- Non-MPAI members who are Respondents.
- Non-Respondents/non-MPAI member experts invited in a consulting capacity.
- (No one from 1.1.-1.2. will be denied membership in the Evaluation Panel).
- Respondents present their proposals to the Evaluation Panel and respond to any questions.
- Evaluation Panel members fill out the Annex 2 form for each proposal.
- A Final Evaluation Report is produced.
The Evaluation Form will contain the following elements.
Proposal title:
Response summary: (a few lines)
Assessment of Response features:
Table 2 – Features to be considered in a Response
| Submission features | Evaluation Elements |
| Completeness of proposal | A detailed description is recommended |
| Understandability of proposal | A detailed description is recommended |
| Implementability of proposal | |
| Complexity of proposal | The implementation complexity and technologies that are potentially usable to implement the proposal. |
| Reliance on established technologies | When assessments are similar, preference will be given to proven technologies. |
| Existence of Essential Patents and Copyright | A proposal that is unencumbered by IP may be selected against a functionally equivalent encumbered proposal. |
9 Timeline
Timeline of the Call, deadlines and response evaluation:
Table 4 – Dates and deadlines
| Step | Deadline | Time |
| Call issued | 2026/08/19 | 17:00 UTC |
| Online presentation | 2026/09/10 | 15:00 UTC |
| 2026/09/10 | 08:00 UTC | |
| Notification of intention to submit a Response | 2026/09/27 | 23.59 UTC |
| Imperative deadline for submission of Responses | 2026/10/27 | 15.00 UTC |
10 References
- MPAI; The MPAI Statutes
- MPAI; The MPAI Standards
- MPAI; The MPAI Workplan
- MPAI; The MPAI Patent Policy
- MPAI; Framework Licence: AI Passport V1.0
- MPAI; Template for Responses; AI Passport V1.0
- MPAI; MPAI Presentations; https://mpai.community/community/presentations/
- MPAI; Technical Specification: AI Framework (MPAI-AIF) V3.0
Annex 1 – Information Form
This information form is to be filled in by a Respondent to the MPAI-AIP Call
- Title of proposal:
- Organisation: company name, position, e-mail of contact person
- What are the main aspects of your proposal?
- Will you provide a demonstration to show how your proposal meets the evaluation criteria?
Annex 2 – Mandatory text in responses
MPAI is willing to consider Responses whose implementation requires exploitation of third-party Intellectual Property. Therefore, even for responses not containing references to proprietary technologies, Respondents are required to include the following statements where Company/Member term is replaced by the name of the Responding non-MPAI member or MPAI member.
A Response to this MaaS Call must include the following text:
<Company/Member> submits this technical document as a Response to Call for Technologies: MPAI as a Service V1.0 (N2880).
<Company/Member> explicitly agrees to the steps of the MPAI standards development process defined in Annex 1 to the MPAI Statutes (N2774), in particular <Company/Member> declares that <Company/Member> or its successors will make available the terms of the Licence related to its Essential Patents according to Framework Licence: MPAI as a Service V1.0 (N2881), alone or jointly with other IPR holders after the approval of the corresponding Technical Specification (“Standard”) by the General Assembly and in no event after commercial implementations of the Technical Specification become available on the market.
In case the Respondent is a non-MPAI member, the submission must include the following text:
If (a part of) this submission is identified for inclusion in an MPAI Technical Specification, <Company> understands that <Company> will be requested to immediately join MPAI and that, if <Company> elects not to join MPAI, this submission will be discarded.
During the development of the Standard, technical contribution must include this text:
<Member> submits this document to MPAI-AIF Development Committee (AIF-DC) as a contribution to the development of a Technical Specification related to MPAI as a Service.
<Member> explicitly agrees to the steps of the MPAI standards development process defined in Annex 1 to the MPAI Statutes (N2774), in particular <Company> declares that <Company> or its successors will make available the terms of the Licence related to its Essential Patents according to Framework Licence: MPAI as a Service V1.0 (N2881), alone or jointly with other IPR holders after the approval of the Technical Specification by the General Assembly and in no event after commercial implementations of the Technical Specification become available on the market.
MPAI is willing to consider Responses whose implementation requires exploitation of third-party Intellectual Property. Therefore, even for responses not containing references to proprietary technologies, Respondents are required to include the following statements where Company/Member term is replaced by the name of the Responding non-MPAI member or MPAI member.
A Response to this MaaS Call must include the following text:
<Company/Member> submits this technical document as a Response to Call for Technologies: MPAI as a Service V1.0 (N2880).
<Company/Member> explicitly agrees to the steps of the MPAI standards development process defined in Annex 1 to the MPAI Statutes (N2774), in particular <Company/Member> declares that <Company/Member> or its successors will make available the terms of the Licence related to its Essential Patents according to Framework Licence: MPAI as a Service V1.0 (N2881), alone or jointly with other IPR holders after the approval of the corresponding Technical Specification (“Standard”) by the General Assembly and in no event after commercial implementations of the Technical Specification become available on the market.
In case the Respondent is a non-MPAI member, the submission must include the following text:
If (a part of) this submission is identified for inclusion in an MPAI Technical Specification, <Company> understands that <Company> will be requested to immediately join MPAI and that, if <Company> elects not to join MPAI, this submission will be discarded.
During the development of the Standard, technical contribution must include this text:
<Member> submits this document to MPAI-AIF Development Committee (AIF-DC) as a contribution to the development of a Technical Specification related to MPAI as a Service.
<Member> explicitly agrees to the steps of the MPAI standards development process defined in Annex 1 to the MPAI Statutes (N2774), in particular <Company> declares that <Company> or its successors will make available the terms of the Licence related to its Essential Patents according to Framework Licence: MPAI as a Service V1.0 (N2881), alone or jointly with other IPR holders after the approval of the Technical Specification by the General Assembly and in no event after commercial implementations of the Technical Specification become available on the market.
Annex 3 – MPAI standards relevant to the Call
A3.1 Data Type identification
MPAI standards currently identify data types with the following identification system
^[A-Z]{3}-[A-Z]{3}-V\d+\.\d+$
A3.2 Data Qualifiers
A block of data is characterised by a string of arbitrary length and a Qualifier specifying the syntax of the data. A Qualifier contains
- Sub Data Types
- Formats for content and transport
- Attribute for other characteristics
And is is expressed by a JSON object conforming to the specific Qualifier.
We should have only Interface without distinguishing between input and output
