Implementation

The Liccium Platform

The FAIA attribution framework is integrated into the Liccium platform as a plugin. This enables creators, researchers, publishers, and platforms to declare AI involvement directly within their content authentication workflows. Designed to be intuitive yet technically robust, the plugin supports transparency, compliance, and interoperability by making AI contributions to content creation verifiable and persistently linked to the original work.

Liccium application

How does it work?

1. Select the Type of AI Involvement

Users can choose from a set of standardised Attribution Flags that reflect the degree and nature of AI’s contribution. These include categories such as:

• Human-Created Content (HCC)

• AI-Assisted Content (AAC)

• AI-Generated Content (AIG)

• Human-Edited AI Content (HEAIG)

• AI-Supported Editorial Processes (AISEP)

This classification ensures consistent disclosure across diverse media formats and sectors.

Be aware that the framework is still in development!

2. Provide Contextual Metadata

Beyond the basic attribution flag, users may optionally provide additional details that clarify how AI was involved, including:

• The name and version of the AI tool (e.g. ChatGPT-4, Midjourney v6)

• The specific function it served (e.g. summarisation, translation, image enhancement)

• Workflow context or editorial stage (e.g. used in initial drafting, applied during peer review)

These metadata elements improve transparency and support reuse, audit, and regulatory evaluation.

Be aware that the framework is still in development!

3. Bind to the ISCC Code

All FAIA declarations are cryptographically bound to the International Standard Content Code (ISCC), a content-derived identifier and ISO standard (ISO 24138). This ensures that attribution metadata remains uniquely and persistently associated with the content – even if the asset is modified and redistributed.

4. Generate Verifiable, Timestamped Declarations

Every declaration is digitally signed using Verifiable Credential (VC) or DID keys of the creator. The declaration is stored as a tamper-evident JSON-LD file (nanopublication), including a trusted timestamp. This guarantees the authenticity and integrity of the AI attribution data.

5. Register in the Liccium Registry

The declarations are published to a public, decentralised registry operated by Liccium, making them:

• Publicly resolvable using content identifiers like the ISCC

• Searchable for researchers, platforms, and compliance tools

• Auditable under regulatory regimes such as the EU AI Act (e.g. Article 50 disclosure obligations)

This registry ensures lasting transparency and verifiability in content attribution and provenance.