This week Several rules of the EU’s law have entered into forcein force since August, focused mainly on the prohibition of AI systems that can be harmful or harmful to citizens, as well as ensuring that whoever has to manipulate and use IAS in certain environments and assumptions has a sufficient level to do them.
There are therefore certain cases of use of AI, including those that involve use to manipulate the behavior of minors, or to cause damage to adolescents. In addition, among the IAS that are already prohibited in the EU are the social score systems that use artificial intelligence, since they can cause unfair and/or disproportionate damage.
So will the use of IAS for assessing the risk in the prediction of criminal behaviors based solely on the creation of profiles. The members of the security forces of the EU countries cannot use IAS to perform biometric identifications in real time in public spaces.
The IAS that take advantage of age, disability or the socio -economic status of people, nor those who try to predict the possibility that people can commit crimes based on their aspect, or those who use the Biometry to “deduce” certain characteristics of a person. For example, your sexual orientation. On the other hand, the IAS that generate or expand facial recognition databases are prohibited by extracting images of the web or security cameras.
Thus, for example, the financial institutions that use the AI for certain processes will have to ensure that their credit risk assessments are not social assessment systems promoted by artificial intelligence. The companies of the European Union, or those of other countries but operate in the region, will have to comply with these standards. Otherwise, they face fines that can be large
In addition to not being able to use AI systems for these mentioned purposes, companies that offer AI systems, or use them, will have to have personnel who have a sufficient level of training in AI. This may achieve it through internal training or by hiring personnel with specific skills for it.
The prohibitions that have just entered into the EU represent only a portion of the measures reflected in the EU’s law. Next month of April, the European Commission will also publish the final version of the Practices Code for AI models General purpose. The measures of this code will enter into force in August of this year. Also then the powers of the supervisory authorities of the Member States of the European Union will be expanded, so that they can ensure compliance with the norms included in the law.
