The European Parliament gives the green light to a binding law regulating the use of AI, placing fundamental rights and transparency at the centre.
During an extraordinary session the European Parliament voted in favor of the regulation on artificial intelligence: a significant step towards regulating the use of technology in the European Union. With a large margin of favorable votes, the regulation, the result of an agreement between Member States, was enthusiastically welcomed as an important breakthrough in protecting citizens' rights and promoting responsible innovation.
The rules, approved with 523 votes in favour, 46 against and 49 abstentionsaims to establish clear guidelines for the use of AI, mitigating any risks and ensuring that the technology respects fundamental European values, such as democracy, the rule of law and environmental sustainability. This legislation represents a major step forward in global technology regulation, placing the European Union at the forefront of ensuring the responsible use of AI.
New laws, fines and bans
Among the key provisions of the regulation is the ban on several AI applications that threaten citizens' rights, such as indiscriminate facial recognition systems, the categorization of sensitive features and emotion monitoring in the workplace and schools. Furthermore, they are introduced Clear obligations for high-risk AI systemsincluding risk assessments, transparency and human surveillance.
Predictive policing practices are also prohibited, if based solely on profiling or the evaluation of a person's characteristics. These dynamics, however, are managed with exceptions: this then opens up if real-time identification is done with rigorous guaranteesperhaps with judicial or administrative approval.
Generative models, according to this AI Act, will have to comply with EU copyright rules during the training phases: according to what has been said, what the various models will be taught during this phase must comply with certain transparency requirements.
Finally, for the most powerful models, those that could really create problems and risks, there will be other obligations to comply with, including assessments of the models and possible systemic risks, complete with communication regarding incidents.
Now it is the turn of the lawyer-linguists, and only after this step can the law be adopted by the end of the legislature. Regardless, it should be highlighted that the law has not been formally approved by the European Council; furthermore, it will only come into force 20 days after publication in the Official Journal of the EU, and will be applicable after 2 years (3 in the case of high-risk models, while only 6 months for generative models).
#Act #Approved #European #Parliament #dictates #law #artificial #intelligence #fines #bans