On June 14, 2023, the European Parliament adopted its negotiating position on the proposed AI Act. This was a step toward negotiations with other EU institutions, not final enactment of the law. The proposal sought safeguards for health, safety, fundamental rights, democracy and the environment. This article preserves the June 2023 policy discussion, not current compliance guidance.
However, the approach taken in the proposal sparked controversy. While it aims to regulate the impact and use of AI, it also extends its reach to the technology itself, including foundational or generative AI. This comprehensive regulation has raised concerns among some, who fear that it might impede innovation.
The rapid uptake of generative AI tools such as ChatGPT brought additional attention to the debate. The following discussion draws on Nuria Talayero and Paloma Villa Mateos’s June 16, 2023 Telefónica commentary. Its regulatory recommendations are the company’s policy position, not findings that a particular governance model guarantees safety.
CEOs from the AI industry, such as Sam Altman from OpenAI and Demis Hassabis from Google DeepMind, were signatories to the statement emphasizing the importance of prioritizing efforts to prevent AI-related risks worldwide, along with other significant global risks like nuclear war and pandemics.
There are different opinions on whether AI can save the world. Marc Andreessen, a co-creator of Mosaic, argued for AI’s potential benefits in his June 6, 2023 essay “Why AI Will Save the World”. This was an optimistic opinion, not proof of future outcomes. Mosaic was not the first web browser; CERN’s history describes Tim Berners-Lee’s earlier browser. Although many acknowledge the importance of managing the risks associated with AI, advocates caution against implementing strict regulations or outright bans. They believe that the fast development of technology will outpace regulatory measures, which could cause harm to both society and the economy.
The three pillars of governance
The retained GPTOnline cover and three-pillar graphic are editorial illustrations of the June 2023 discussion. “Global guidelines,” “Self-regulation” and “Regulatory framework” describe the proposed governance framing, not a list of legal requirements.
The focus is on promoting responsible digitalization as a fundamental strategy in guiding the creation and application of technology, including generative AI. As an example, companies such as Telefónica have adopted an AI governance model that incorporates ethical principles and transparency mandates, through a “Responsibility by Design” approach.
In the June 2023 commentary, Telefónica argued that EU institutions should consider how AI can transform the economy and society positively. AI is a technology that can improve competitiveness in both regional and business contexts. Therefore, the AI Act that results from this should encourage innovation and help the internal market function better.
Telefónica presented AI as a potential contributor to business competitiveness. It can lead to the creation of new business models, innovative services, improved efficiency in operations, and positive social impact. However, it is important to approach innovation in a human-centric and trustworthy manner. Telefónica’s public statement demonstrates its dedication to enhancing prosperity while protecting people’s rights and preserving societal values.
Telefónica proposed a governance framework with three parts. These three parts are global guidelines, self-regulation, and a regulatory framework that is appropriate. It is recommended that the EU AI Act have a narrow focus and be complemented with regional guidelines that comply with worldwide agreements and ethical principles upheld by both public and private stakeholders.
The commentary argued that national rules alone would be insufficient for cross-border AI governance. This means we need international cooperation and guidelines to make sure everyone follows the same ethical principles and best practices. It pointed to then-emerging efforts toward voluntary codes of conduct and non-binding international standards, such as the AI Pact in Europe and collaborations between the EU and the US.
Self-regulation can be beneficial for AI development, particularly for non-high-risk applications. It can offer flexibility and efficiency in regulating innovation without hindering progress. Telefónica argued that self-regulation could complement protections for rights, health, democracy and safety; the proposal did not establish that voluntary measures alone ensure those outcomes. A priori regulations might not be effective due to the fast pace of innovation and the complex nature of AI.
Telefónica has adopted ethical principles across all its operations, from design and development to the use of AI products and services by employees, suppliers, and third parties. The company described its “Responsibility by Design” approach as intended to incorporate ethical and sustainable criteria throughout the value chain.
To sum up, regulating AI requires a broad approach that includes global teamwork, self-regulation, good public policies, and a regulatory system based on risk. This comprehensive strategy intends to reduce risks, promote ethical technology use, and create a strong foundation for innovation and economic progress. By finding the right balance between innovation, ethics, and regulations, we can shape a future where AI empowers us while respecting the values and rights of our society.
