Artificial Intelligence Find out what general purpose AI models and systems are

ANACOM plays a central role in applying the Artificial Intelligence Acthttps://www.anacom.pt/render.jsp?contentId=1790431(AI Act) as one of the bodies designated under its article 77 to protect fundamental rights, and is concerned with promoting AI digital literacy.

To this end, ANACOM is publishing a set of FAQs on general-purpose AI models and systems within the scope of the AI Act:

1. What are general-purpose AI models?

General-purpose artificial intelligence (AI) models are AI models that are trained with a large amount of data, have significant generality, and can be used for a wide range of different tasks, such as speech/image recognition, audio/video generation, or question answering. General-purpose AI models can be brought to market in a variety of ways, such as libraries or application programming interfaces. These models can be modified or refined into new models.

2. Are general-purpose AI models a subset of AI systems?

No. AI models are essential components of AI systems, but they do not constitute AI systems by themselves. AI models are typically integrated and form an integral part of AI systems, requiring the addition of other components - such as a user interface - to become AI systems.

3. What are general-purpose AI systems?

General-purpose AI systems are AI systems that are based on general-purpose AI models and therefore have the ability to be used for a variety of functions. General-purpose AI systems can be used directly or as components of other systems or applications.

4. Which AI systems can be classified as general-purpose AI systems?

AI models that can be classified as general-purpose AI models are, for example, the large generative AI models that allow the creation of new original content in different formats (e.g. image, video, audio). AI systems based on such models are therefore general-purpose AI systems, such as OpenAI's ChatGPT, Google's Gemini and Microsoft's Copilot.

5. Do general-purpose AI models fall under the Artificial Intelligence Act?

Yes.

6. What are the obligations of providers of general-purpose AI models under the AI Act?

The AI Acthttps://www.anacom.pt/render.jsp?contentId=1790431provides specific rules for general-purpose AI models, in particular with regard to transparency and copyright. Where general-purpose AI models pose systemic risks, the Act provides for additional obligations to ensure that these risks are identified, assessed and mitigated. It should be noted that the specific rules for general-purpose AI models and for general-purpose AI models that pose systemic risks should also apply when these models are integrated or form an integral part of an AI system.

7. What is systemic risk?

A systemic risk is a risk specific to general-purpose AI models that has a significant impact because of its reach or because of its negative impact on public health, safety, security, fundamental rights or society, and that can spread and scale along the value chain.

8. When will the AI Act rules on general-purpose AI models come into force?

The AI Acthttps://www.anacom.pt/render.jsp?contentId=1790431rules on general-purpose AI models will come into force on 2 August 2025.

9. What is the General-Purpose AI Code of Conduct?

The General-Purpose AI Code of Conduct, which is currently being drafted, will set out the rules of the AI Act that apply to providers of general-purpose AI models and general-purpose AI models that pose systemic risks. Various stakeholders are involved in the drafting of the Code, including providers of general-purpose AI models, civil society and other organisations. The third draft of the Code https://digital-strategy.ec.europa.eu/pt/library/third-draft-general-purpose-ai-code-practice-published-written-independent-expertswas published by the European Commission on 11 March 2025 and is expected to be finalised in May 2025.