EU AI Act
The European Parliament approved the Artificial Intelligence Act with overwhelming support, marking a significant milestone as the first major political entity to regulate AI. The main goal is to secure both European innovation and human rights. The act prohibits certain AI applications like algorithmic social scoring and exploitative practices while imposing strict regulations on high-risk AI systems used in areas like education and work. This is going to affect all companies, people and governments operating in Europe. So what is at stake here, from an educational perspective?
First of all, it depends on how the law will be applied, which is not clear yet. The EEA has warned about the risks of poor implementation and advised not to make the same mistake as with GDPR, leaving SMEs and other actors without any guidance. It is too soon to conclude the AI Act and its implications, but we would like to set some topics in the limelight:
Flexible law:
We see a high degree of flexibility in the law and it will be interesting how the law can follow the speed at which technology develops. This is positive from one perspective as technology develops faster than ever, but from an investor perspective, and don't forget, this is a technology demanding high-level investment to be competitive, this can also be a downturn in AI investments in European companies. The high risk could be that the regulations change over time, good or bad.
Small and medium enterprises: If demands on European SMEs are too high to adhere to the regulatory requirements, we will “have created a monster” some are saying, if investment and compliance levels are too high, we will leave walkover to third-country companies and the European competitiveness are in danger.
The AI Act imposes significant penalties for non-compliance with the prohibited systems provisions, with fines of up to €35 million or 7% of global turnover.
General purpose AI:
The companies behind generative AI must provide a detailed summary of what images, videos and other data they trained the AI systems with. Those made up of real people, places and events must also be flagged as having been artificially manipulated.
Advanced AI models labelled "general-purpose" must adhere to specific mitigation and testing obligations, enforced by a newly established unit within the European Commission called the AI Office.
What the AI Act explicitly says about education:
Emotion Inference in Workplaces and Educational Institutions: The use of AI to infer emotions in sensitive environments like workplaces and schools is banned, barring exceptions for medical or safety reasons.
Exploitation of Vulnerabilities: The Act prohibits AI systems that target individuals or groups based on age, disability, or socio-economic status to distort behaviour in harmful ways.
Further Reading:
https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai