EU Parliament agrees on the "Artificial Intelligence Act"
As the first international institution, the European Parliament has reached an agreement on a legislative initiative intended to regulate and control artificial intelligence (AI), known as the “Artificial Intelligence Act”.
A preliminary compromise by the EU Parliament on the so-called “Artificial Intelligence Act” aims to ensure that artificial intelligence (AI) in Europe is safe, respects fundamental rights and democracy, while simultaneously allowing businesses to grow and expand. The framework aims to protect fundamental rights, democracy, the rule of law, and environmental sustainability from high-risk AI, while also promoting innovation and making Europe a leading player in the field of AI. The “Artificial Intelligence Act” establishes clear obligations for AI systems based on their potential risk and impact level.
Criticism in Germany came from the business sector and the FDP, with concerns that the legislative proposal might be too stringent in its requirements, potentially hindering innovation and economic growth. Similar criticism was voiced in other European countries. The French government also pushed for more business-friendly solutions.
Key Elements of the Regulation
- Rules for highly effective general AI models that could cause systemic risk in the future, as well as for high-risk AI systems
- A revised governance system with some enforcement powers at the EU level
- Expansion of the prohibition list, but with the possibility for law enforcement agencies to use biometric identification in public spaces under very specific circumstances
- Enhanced protection of rights through the obligation for operators of high-risk AI systems to assess the impact on fundamental rights before deployment
Prohibited Applications
According to the AI Act, certain actions by unregulated AI are in direct conflict with EU values and laws. Therefore, these actions will be explicitly prohibited to protect citizens’ rights and democracy.
- Categorization systems that use sensitive features such as political, religious or philosophical beliefs, as well as sexual orientation or race, will be prohibited
- The indiscriminate collection of facial images for the creation of facial recognition databases
- Emotion recognition at the workplace and in educational institutions
- Social scoring of individuals based on social behavior or personal characteristics
- AI systems that manipulate human behavior to circumvent free will and manipulate decisions
- AI exploiting human sensitivities (due to age, disabilities, social or economic situation)
Exceptions for Law Enforcement Authorities
Protective measures and strictly limited exceptions for the use of identification systems in publicly accessible spaces for law enforcement purposes have been agreed upon. These are subject to prior judicial authorization and are limited to specifically defined offenses, such as suspected terrorist attacks or the search for major criminals.
Obligations for High-Risk Systems
Clear obligations have been agreed upon for AI systems classified as high-risk. These include a mandatory assessment of the impact on fundamental rights, applicable also to the insurance and banking sectors. AI systems used to influence election outcomes and voter behavior are also considered high-risk. Citizens have the right to file complaints about AI systems and receive explanations about decisions based on high-risk AI systems that affect their rights.
Safeguards for General AI Systems
General AI systems and the models they are based on must meet precise transparency requirements. This includes the creation of technical documentation, compliance with EU copyright laws, and the dissemination of detailed summaries of the content used for training.
Sanctions and Enforcement
Depending on the nature of the violation and the size of the company, non-compliance can lead to fines ranging from 1.5-7% of the company’s global turnover. With the EU Parliament’s consent, member states now have two years to gradually decommission all prohibited systems. By 2026, all points of the AI Act are expected to be implemented.
New Governance Architecture
An agreement was also reached on establishing an AI office within the Commission to monitor the most advanced AI models. An AI Board, consisting of representatives from the member states, will remain as a coordination platform and advisory body for the Commission.
Implications for Businesses
Companies that use AI must be prepared for new risks regarding regulatory violations. Most compliance programs are not yet designed to address the challenges posed by AI systems. For the AI Act, companies will need to revise their programs with the AI alignment problem in mind to detect and prevent AI-related regulatory violations. This requires special training programs for employees to inform them about the compliance challenges posed by AI. Monitoring AI-based behavior using surveillance AI systems that can detect anomalies and illegal behavior and automatically trigger alarms might be one option.
If you have any questions, Compliance Solutions will be happy to assist you with its many years of expertise
Conclusion
The EU’s “Artificial Intelligence Act” is a significant legislative initiative aimed at comprehensively regulating artificial intelligence (AI) to ensure it is safe, respects fundamental rights and democracy, while also supporting the growth and expansion of businesses. By establishing obligations for AI systems based on their risk and impact level, the legislation aims to minimize risks and promote innovation, making Europe a leading player in the AI field. The proposed measures include strict rules for high-risk and general AI systems, a revised governance structure, and specific prohibitions against abusive practices, with certain exceptions granted for law enforcement authorities under strictly defined conditions. For companies, there is a new complex of risks that will have an impact on compliance departments in order to build the skills to get to grips with the AI alignment problem. It is now up to the member states to implement the AI Act conscientiously and introduce appropriate sanctions against non-compliance.
