In this article7 sections
  1. What Article 2(12) Actually Says
  2. Who the Exemption Is Actually For
  3. When Fine-Tuning Takes You Out of the Exemption
  4. What About Unmodified Deployment?
  5. Practical Implications for Mistral, Llama, and Similar Models
  6. The Deadline Picture
  7. References

The EU AI Act open source models exemption sounds like a clean escape hatch. It is not. Article 2(12) of the AI Act carves out AI systems released under free and open-source licences, but only as long as they are not used for high-risk purposes, prohibited practices or anything covered by the transparency rules. The moment your company builds one of those models into a business system that falls in one of those categories, the exemption is gone. This post explains exactly where the line sits, when you cross it, and what that means for your compliance posture today.

What Article 2(12) Actually Says

The AI Act's open source carve-out for AI systems is found in Article 2(12). In short, the Regulation 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 AI systems, or as systems that fall under Article 5 (prohibited practices) or Article 50 (transparency).

There is a second, separate carve-out for general-purpose AI models. Under Article 53, providers of models released under a free and open-source licence, with their parameters made publicly available, are relieved of some documentation duties toward the AI Office and downstream providers. That relief does not apply to models with systemic risk, and duties such as a copyright policy and a public summary of training content still apply.

So the exemptions are about releasing systems and models openly. They protect open source developers, researchers, and communities from bearing the same compliance burden as a commercial AI company shipping a finished product. Think of the teams publishing foundational weights on Hugging Face rather than a company building a payroll-screening tool on top of Llama 3. They do not grant a blanket exemption across the entire regulation.

The recitals of the AI Act explain the intent: open source release is encouraged because it supports innovation and research, but that encouragement stops where high-risk use begins.

Who the Exemption Is Actually For

Before going further, it helps to be clear about who sits at each point in the supply chain.

  • Original developers publishing model weights openly (e.g., Meta releasing Llama, Mistral AI releasing Mistral 7B) may benefit from the open source carve-out at the point of release.

  • Deployers, meaning companies that take those weights and use them to build or run an AI system in their own business, do not automatically inherit that exemption.

This matters enormously. As a deployer under Article 26, your obligations flow from how you use the system, not from the licence the underlying model carries. Open source does not mean obligation-free when it lands inside your HR platform, your customer service bot, or your document screening tool.

When Fine-Tuning Takes You Out of the Exemption

This is where things get concrete. Suppose your company takes an open source model, say Mistral 7B or Llama 3 8B, and fine-tunes it on your own HR data to assist with CV screening. The original model release may have qualified for the open source carve-out. Your fine-tuned derivative almost certainly does not.

Here is why. When you build an AI system on top of a model and put it into service under your own name, you are the provider of that system under Article 3(3) of the AI Act, not merely a deployer. And because CV screening is a high-risk use, the open source carve-out in Article 2(12) does not apply to it. Provider obligations then apply, and they are substantially heavier.

For high-risk systems, from 2 December 2027 for Annex III uses, those obligations include:

  • Establishing a quality management system (Article 17), in a simplified form for many SMEs (Article 63)

  • Conducting a conformity assessment before putting the system into service (Article 43)

  • Registering the system in the EU database (Article 49)

  • Drawing up technical documentation (Article 11 and Annex IV), which SMEs and small mid-caps may provide in a simplified form

CV screening and recruitment tools fall squarely into Annex III as high-risk AI systems. Fine-tuning an open source model for that purpose does not reduce the regulatory classification. It may actually deepen it.

What About Unmodified Deployment?

Suppose you do not fine-tune at all. You take a publicly released model and deploy it as-is through an API wrapper or on your own infrastructure. Does the exemption protect you then?

No. Even unmodified use for a business purpose makes you a deployer under the Act. A system you run inside your own business is not one "released under a free and open-source licence", so Article 4 (AI literacy) applies to you whatever the model's licence. If you use the system for a high-risk purpose, the deployer obligations under Article 26 apply from 2 December 2027 regardless of the model's open source status. Those include:

  • Using the system according to the instructions for use

  • Assigning human oversight to competent, trained staff

  • Monitoring the system and, if it presents a risk or causes a serious incident, informing the provider and the market surveillance authority (Article 26(5))

  • Keeping the logs the system generates for at least six months

  • Informing workers, and people subject to decisions, that a high-risk system is used (Article 26(7) and (11))

  • A Fundamental Rights Impact Assessment (Article 27), but only if you are a public body, a private provider of public services, or you use the system for credit scoring or life and health insurance pricing

If you wire an open model into a high-risk use yourself, check first whether that makes you the provider of the resulting system; that question is not always clear-cut, and the answer decides which list applies.

The open source label on the model's licence file changes none of this. Your legal obligation is determined by what the system does and who is affected by it.

Practical Implications for Mistral, Llama, and Similar Models

Let us get specific about the models European SMEs are actually using.

Mistral models (Mistral 7B, Mixtral, Mistral Small) are released under the Apache 2.0 licence. The company behind them is a commercial entity. Whether any open source carve-out helps Mistral AI as the model provider is its question, not yours. For your own high-risk or customer-facing system, it does not.

Meta's Llama models (Llama 2, Llama 3, Llama 3.1) use a custom community licence. Again, the release may benefit the original publisher. Your deployment does not.

Practical checklist for any SME using these models:

  1. Identify the use case. Is the system making or supporting decisions about people in employment, credit, education, or essential services? If yes, it is likely high-risk under Annex III.

  2. Determine your role. Did you build the system yourself, with or without fine-tuning, and put it into service under your own name? If yes, you are likely its provider, not just a deployer.

  3. Check transparency obligations. If you build a chatbot on an open model, Article 50 requires that people are told they are interacting with AI, and synthetic text, audio or images must be marked as AI-generated in a machine-readable way. As the builder, these are your duties as provider. If you publish deepfakes, you must disclose them as a deployer. This applies whether the underlying model is open source or proprietary.

  4. Support AI literacy. Article 4, as amended by Regulation (EU) 2026/1744, requires you to take measures to support the AI literacy of staff who work with AI systems, including how those systems work and their limitations. No specific level per person is required. Open source models are not exempt from this obligation.

  5. Document everything. Whether you are a provider or deployer, the AI Act rewards documented decision-making. Record why you chose the model, how you tested it, and what risk controls you applied.

The Deadline Picture

The prohibitions (Article 5) and AI literacy (Article 4) have applied since 2 February 2025; two new prohibitions (non-consensual sexual deepfakes and child sexual abuse material) apply from 2 December 2026. General-purpose AI model obligations (Chapter V) apply from 2 August 2025 for providers of those models. The transparency rules in Article 50 apply from 2 August 2026. After the Digital Omnibus on AI, obligations for high-risk AI systems in Annex III start applying on 2 December 2027, and for AI in products covered by Annex I on 2 August 2028.

If you use an open source model in a high-risk context, December 2027 is your preparation horizon. Conformity assessments, log management systems, and oversight processes take months to implement properly, especially for companies without a dedicated compliance function. Taking AI literacy measures for your staff is already required today.

Start the inventory now. Map every AI system in use, note the underlying model, and assign a risk classification. That single step, done rigorously, will tell you whether the open source label on your model actually changes anything for your compliance obligations. Spoiler: in most real business deployments, it does not change much at all.

Run your free 2-minute compliance check at comply.khairos.ai to see where your current AI systems sit on the risk scale and which obligations already apply to your business.

References

  1. Article 2(12)

  2. open source release

  3. Article 26

  4. Annex III

  5. Article 53

  6. Regulation (EU) 2026/1744 (Digital Omnibus on AI)

  7. comply.khairos.ai

Not sure where your company stands?

Answer 15 questions about how your team uses AI. You get a compliance score and the gaps to close first, as a PDF.

Am I compliant? Check in 2 minHow Khairos Comply works