# EOLE 2024 synthesis: AI Act & Open Source

**URL:** https://forum.eolevent.eu/t/eole-2024-synthesis-ai-act-open-source/98
**Category:** AI Act
**Created:** [22 June 2026 09:13 UTC](https://forum.eolevent.eu/t/eole-2024-synthesis-ai-act-open-source/98 "2026-06-22T09:13:12Z")
**Posts on this page:** 1
**Page:** 1

<div class="post-metadata">

### Author: ![root](https://forum.eolevent.eu/user_avatar/forum.eolevent.eu/root/32/9_2.png) [@root](https://forum.eolevent.eu/u/root)
#### Post date: [22 June 2026 09:13 UTC](https://forum.eolevent.eu/t/eole-2024-synthesis-ai-act-open-source/98/1 "2026-06-22T09:13:12Z")

</div>

Summary of the **AI Act** session at EOLE 2024, part of _Navigating the New EU Regulatory Landscape in Open Source_.

**Context.** The AI Act provides exceptions for “free and open-source” AI, but the meaning of a “free and open-source AI system” remains uncertain, and recent developments leave room for interpretation.

**Risk-based approach.** The AI Act does not apply to AI systems released under free and open-source licences, unless they are placed on the market or put into service as high-risk systems, or fall under Article 5 (prohibited practices) or Article 50 (e.g. chatbots, deep fakes).

**Exceptions for general-purpose AI (GPAI) model providers.** Article 25(4) for third-party FOSS tools and components, Article 53(2), and Article 54(6). Recital 104 notes that releasing a GPAI model under a free and open-source licence does not necessarily reveal substantial information about the training data.

**OSI’s Open Source AI Definition.** AI models have no “source code” in the traditional sense, so OSI reinterpreted the four freedoms for AI (use, study, modify, share). The definition ([Open Source AI – Open Source Initiative](http://opensource.org/ai)) is endorsed by EleutherAI, LLM360, Software Heritage, Eclipse and Mozilla, and contrasts with the Free Software Foundation’s position, a productive and ongoing debate. Restrictive licences such as Meta’s Llama (field-of-use limits) fall outside Open Source.

**Public administrations.** AI must serve the common good: equality and constitutional rights, transparency, and the diligence of a good administrator. Transparency alone is not enough (systems can be transparent yet remain black boxes). A “public-oriented”, controlled AI, combined with open data and open AI, is key to preserving sovereignty.

**Resources**

Marco Ciurcina, Free and open source exceptions in the AI Act: [Free and open source exceptions in AI Act – EOLE](https://eolevent.eu/free-and-open-source-exceptions-in-ai-act/)

Laura Garbati, AI and transparency in public administrations: [Artificial Intelligence & OS : AI and transparency in Public Administrations – EOLE](https://eolevent.eu/artificial-intelligence-os-ai-and-transparency-in-public-administrations/)

Carlo Piana, The Open Source AI Definition: [The Open Source AI Definition – EOLE](https://eolevent.eu/the-open-source-ai-definition/)

Full EOLE 2024 synthesis: [EOLE 2024 – Navigating the New EU Regulatory Landscape in Open Source – EOLE](https://eolevent.eu/eole-2024-navigating-the-new-eu-regulatory-landscape-in-open-source/)

What’s your reading of the “open source AI” debate? Share your thoughts below.