Position Paper on Artificial Intelligence

As our world continues to be shaped by unprecedented advancements in technology, particularly in the

This statement is a testament to EuroISPA’s collective dedication to shaping a future where AI serves as a
force for good. It reflects our commitment to establishing a harmonised and globally accepted framework
that encourages innovation, upholds ethical standards, and mitigates potential risks associated with AI
deployment.

Below you can find principles that should be considered for current and future regulatory frameworks on
AI, including by the European Union.

1. Copyright and intermediary liability of AI: Promoting balanced copyright regulation that does not
burden operators (ISPs) with excessive liability for AI-generated content. Additionally, owners of
training data should not be able to leverage their exclusive rights over foundation models.
2. Follow a Global Definition of AI: Advocating for the establishment of a universally accepted
definition of AI such as the ones adopted by the OECD and the EU AI Act which provide clarity and
consistency in regulatory frameworks worldwide, ensuring a common understanding of the
technologies covered. The definition must also accurately reflect the special features of AI so as not
to include classic software applications.
3. Balance between Regulation and Innovation: Emphasising the importance of striking a delicate
balance between regulating the application and fostering innovation. Avoiding unnecessary
complexity and bureaucracy is crucial to maintaining an environment where businesses can thrive
while ensuring responsible AI development. One should also think of alternatives to regulation when
relevant, such as the encouragement of Codes of Conduct as they are agile and unbureaucratic
1
alternatives .
4. Rules for Users and SMEs: Promoting regulations that cater to the needs of users and SMEs.
Implementing measures that allow businesses, especially internet service providers to leverage AI
tools without facing high costs or complexity, fostering widespread adoption.
5. Support of a Risk-Based Approach: Acknowledging the necessity of a risk-based approach to AI
regulation, where the level of regulatory scrutiny corresponds to the potential risks associated with
specific AI applications. At the same time, we should try to stay technologically neutral and build

1
Example of a welcomed Code of Conduct on AI is the G7 agreement on international guiding principes for
Organisations Developing Advanced AI Systems under the Hiroshima Process.

EuroISPA
Rue de la Loi 38, 1000 Brussels
[email protected]
EU Transparency Register: 54437813115-56
 future-proof regulation. These considerations allow for a nuanced and proportional regulatory
framework.
6. Human Oversight: Recognising the significance of human oversight in AI systems especially in high-
risk scenarios. Ensuring that there are mechanisms in place to maintain human oversight over critical
decision-making processes, particularly in situations where there is an impact on fundamental rights.
7. Role of National Competent Authorities: Defining the appropriate level of competent authorities
in overseeing AI regulations and enforcement. Clarify the responsibilities and powers of these
authorities to ensure effective governance and harmonised application of standards within the EU
at Member State level.
8. Classification of the Role of Generative AI and Foundation models: Establishing clear
classifications for different types of AI, such as AI foundation models and generative AI, to tailor
(self-)regulatory requirements based on the nature and potential risks associated with each category
and its use cases. A general classification of generative AI and foundation models into a high-risk
category is not appropriate in terms of the risk-based approach.
9. Establishing a unified Regulatory framework: Ensuring that AI-specific regulations do not overlap
with existing legislation (such as the DSA and the GDPR). The goal should be to construct a legal
framework that tackles the distinctive challenges posed by AI while also integrating and aligning
regulations harmoniously, fostering a seamless ecosystem. Additionally, it's crucial to contemplate
the interplay between current horizontal European legislation (AI Act) and potential forthcoming
specialised laws (e.g., copyright, employment) for comprehensive coherence.
10. Infrastructure & environment: AI has the potential to optimise processes, reduce environmental
impact and make more efficient use of resources. This potential must be consistently utilised. In
order to further increase the benefits of AI applications for tackling climate change and
environmental problems, the research and development of energy-efficient algorithms should also
be supported and promoted more strongly.

About EuroISPA

Established in 1997, EuroISPA is the world's largest association of Internet Services Providers Associations,
representing over 3,300 Internet Service Providers (ISPs) across the EU and EFTA countries. EuroISPA is
recognised as the voice of the EU ISP industry, reflecting the views of ISPs of all sizes from across its
member base.

EuroISPA
Rue de la Loi 38, 1000 Brussels
[email protected]
EU Transparency Register: 54437813115-56