The fact that artificial intelligence is – – a double-edged sword, no one today seems to doubt. Deepfake, image generation or mass copyright infringement by AI are just some of the threats. A sample of the possibilities of AI was well shown, even false. Some people actually believed that they were real.
The European Union wants to tame AI. But there is one problem
The remedy is to be the AI Act, i.e. the provisions regulating artificial intelligence on the Old Continent. EU lawmakers have been working on them for several months now and the work is already on the final straight. The regulation on artificial intelligence is to divide it according to the threat: unacceptable risk (prohibited), high risk (strictly regulated), acceptable risk (acceptable under certain restrictions) and minimal risk (allowed for free use). The regulations are to impose, for example, on suppliers of AI tools the obligation to block the generation of illegal content or inform the user that a given image is artificially generated.
One of the technologies whose use is to be severely restricted is remote biometric identification of people, i.e. tracking carried out most often through cameras at stations, shopping centers and other public spaces. According to the previously developed consensus, the recognition and tracking of citizens in real time using this method would be allowed only in situations of searching for specific persons and only with the consent of the court. Panoptykon Foundation, for the European People’s Party (EPP) this turned out to be too far-reaching limitation. This week, just before Wednesday’s vote on the AI Act, the EPP submitted a proposal for amendments related to this technology.
Will this be a gateway to mass surveillance of people?
They show that the ban on biometric identification can be lifted by the government in exceptional situations. According to the foundation, for situations such as searching for a missing person, suspected terrorist threat and searching for persons suspected of committing crimes punishable by imprisonment of more than three years. In practice, there can be quite a lot of such “exceptional situations”. It is easy to guess that the governments of EU countries can use such a loophole and use technology to surveillance citizens during large-scale events, gatherings or even anti-government protests. This carries quite obvious risks. And it is worth adding here that the use of biometric identification by governments is not a domain of the future. This technology has been around for years
Panopticon also reminds us that artificial intelligence is not always infallible. The risk of making a mistake and accusing a random person may not be great, but it exists and it does not apply to everyone to the same extent. In order for the system to correctly recognize human behavior, it must use huge amounts of data on as many people as possible. And not all groups are equally representative in the databases on which AI tools are trained. We will probably find out about the form in which the AI Act will finally come into force on Wednesday.
Source: Gazeta
At 247 News Agency, Jason has covered a wide range of topics, from politics and business to crime and entertainment. He is highly respected by his colleagues and peers, and his articles are widely read and shared. With a passion for news and a commitment to accuracy, Jason Root is a valuable asset to the 247 News Agency team.
