European artificial intelligence act

On August 1, 2024. The European Artificial Intelligence (AI) Act, which is the world’s first comprehensive AI regulation, entered into force. The AI Act imposes clear requirements and obligations on developers and implementers of the tool for specific applications. One of the issues clearly stated in the document is responsibility for resources, including water.

Is an act on artificial intelligence needed?

The uptake of artificial intelligence (AI) systems carries great potential: it can provide social benefits, economic growth, and spur innovation and increase the EU’s global competitiveness. As the use of AI enables better forecasting, optimization of operations and resource allocation, and personalization of digital solutions available to individuals and organizations, it can also provide environmental benefits, such as environmental monitoring, protection and restoration of biodiversity and ecosystems, and climate change mitigation. In some cases, the features of some AI systems can generate new threats to users’ security and basic rights. The development of artificial intelligence also has a particular impact on resources, including water, as we wrote about in a previous article: ChatGPT – a water-intensive giant?

In response to these challenges, the Artificial Intelligence Act was adopted to ensure a well-functioning internal market for AI systems that takes into account both benefits and risks. AI legislation is necessary to promote the development, use and spread of AI in the internal market, while ensuring a high level of protection of public interests, such as health and safety, and fundamental rights, including democracy, the rule of law and environmental protection, as recognized and protected by European Union law.

Who is bound by the artificial intelligence act?

The legal framework contained in the European Act on Artificial Intelligence will apply to both public and private entities, inside and outside the European Union, as long as the AI system in question is marketed in the Union or its use affects its residents.

Risk categories

The AI Act introduces a uniform legal framework in all member states. They are based on a risk-based approach and a definition of AI that takes into account further developments in technology:

• minimal risk: Most artificial intelligence systems, such as recommendation systems and spam filters, fall into this category. They are not subject to the obligations of the Artificial Intelligence Act due to their minimal impact on citizens’ rights and security. Companies can voluntarily adopt additional codes of conduct.
• special transparency risks: Artificial intelligence systems, such as chatbots, must clearly disclose to users that they are interacting with a machine.
• HIGH RISK: Artificial intelligence systems identified as high-risk will have to meet stringent requirements, including risk mitigation, high quality data sets, activity logging, detailed documentation, clear user information, human oversight, and high levels of robustness, accuracy and cyber security.
• unacceptable risks: Artificial intelligence systems deemed a clear threat to people’s fundamental rights will be banned. This includes systems or applications that manipulate human behavior to circumvent users’ free will, such as toys that use voice assistance to encourage minors to engage in dangerous behavior.

To complement this system, the Artificial Intelligence Act also introduces provisions for so-called “general-purpose artificial intelligence models. general-purpose artificial intelligence models, which are highly capable tools designed to perform a wide range of tasks, such as generating text.

European act on artificial intelligence and environmental protection

The purpose of the AI proposal is to counter threats to security and fundamental rights, including the right to a high level of environmental protection. The environment is one of the explicitly mentioned and protected legal interests. In the adopted act, the European Commission was required to ask European standards organizations to develop a document on reporting and documentation processes to improve the performance of artificial intelligence, such as reducing the use of energy and other resources, including water, by a high-risk AI system during its life cycle.

The European Commission was also required to report on its review of progress in the development of standardization documents for the energy-efficient development of general-purpose models, and to assess the need to pass further measures or take action, including of a binding nature.

Timetable for implementation of rules under the European Artificial Intelligence Act

Member states have until August 2, 2025 to. to designate competent national authorities to oversee the application of regulations on artificial intelligence systems and conduct market surveillance activities. The Commission’s Office for Artificial Intelligence. The Commission’s Office for Artificial Intelligence will be the key body implementing the act at the EU level, as well as the body responsible for enforcing regulations on general-purpose artificial intelligence models.

Most of the Artificial Intelligence Act’s provisions will take effect on August 2, 2026. However, bans on artificial intelligence systems deemed to pose unacceptable risks will take effect after just six months, while provisions on so-called “general-purpose models. general-purpose models will apply after 12 months.

To complete the transition period before full implementation, the European Commission has launched the Artificial Intelligence Pact. Under this initiative, AI developers are encouraged to voluntarily accept key obligations under the act’s implementation before the legal deadlines.