AI Passport (MPAI‑AIP) – Enabling AI Systems to Understand Each Other

The AI Passport is an MPAI standard project that intends to provide a standardised description of an AI system, including its functionality, interfaces, data semantics, origin, performance, and compliance status, to enable both humans or AI systems to discover, interpret, and trust other AI systems.

Why AI Needs Passports

Today’s AI systems operate as isolated entities with unclear capabilities, ambiguous or undisclosed data formats, difficult-to-verify compliance, requiring case-by-case effort to integrate them. This prevents efficient reuse, automation of integration, and scalable AI‑to‑AI collaboration.

The AI Passport intends to address this gap by making AI systems self‑describing and machine‑understandable.

What the AI Passport Will Provide

An AI Passport is defined as a standard digital description of an AI system that includes:

  • Identity and Integrity: Unique AI Passport ID and cryptographic fingerprint linking the passport to the AI instance.
  • Function and Purpose: Description of what the AI does and its intended operational use.
  • Interfaces and Data: Input/output data formats and communication protocols.
  • Data semantics and structure: The meaning and format of data.
  • Provenance: Ground model and architecture, training data references, development information.
  • Performance: Generic AI KPIs (e.g., accuracy, AUC) and application‑specific AI KPIs.
  • Compliance and Governance: Certifications (regulations, standards); ethical, social, and environmental data governance.

AI‑to‑AI Understanding

A key innovation of the AI Passport is to enable direct machine‑level interaction between AI systems. With AI Passports, an AI system will be able to:

  • Discover another AI with required capabilities.
  • Assess compatibility  between interfaces, protocols, data semantics.
  • Evaluate trustworthiness through certification, provenance, and governance.
  • Decide whether to interact further.

This supports both offline discovery (marketplaces, repositories) and real‑time interaction (dynamic AI orchestration).

The Importance of Data Semantics

AI systems do not merely exchange data, they also need to be aware of its meaning.

The AI Passport will provide data type identifications, semantic qualifiers describing data content, and references to standard taxonomies thus allowing AI systems to understand what data represents, how it was produced, and how it might be used.

Without this, AI‑to‑AI interaction remains risky and error‑prone.

Benefits for Individuals and Organisations

  • For Users: Discover AI systems with precise capabilities and compare alternatives objectively.
  • For Developers: Describe AI systems in a machine‑interpretable form and reduce integration complexity.
  • For Companies: Avoid duplication of functionality and improve reuse and optimisation of resources.
  • For Regulators: Identify and analyse critical AI systems and support enforcement of AI regulations.
  • Bridging the Gap Between Regulation and Practice: Current AI regulations and governance frameworks provide principles, guidelines, and high‑level requirements. However, they lack practical, machine‑processable mechanisms to apply them.

The AI Passport fills this gap by offering a concrete, operational standard to represent AI systems in a verifiable way.

 A Foundation for AI Ecosystems

With AI Passport, AI systems will become discoverable assets, interactions more automated and reliable, ecosystems more scalable and interoperable, thus enabling a shift from isolated AI applications to cooperating AI systems that understand each other’s capabilities.

A Standard for Trustworthy AI Interaction

The AI Passport will define an essential mechanism for AI‑to‑AI interoperability, data understanding and compatibility, and for transparent and trustworthy AI ecosystems. It is a key step toward a future where AI systems can autonomously collaborate based on shared understanding of capabilities, data, and context.

MPAI plans on issuing a Call for AI Passport Technologies on 18 August 2026. Responses to the Call will be due on 26 October 2026.