European Union (EU) completed the first law to regulate the world Artificial intelligence (AI). On June 14, the European Parliament set out its position and gave the green light to a pioneering measure whose mission is to control the spread of this emerging technology. technologysetting more or less stringent standards depending on the level risk Its use requires this. Although it has a large majority, the approved text has not yet been finalized. Its final version will depend on negotiations with the EU. Council of EuropeA decision by the heads of State or Government of 27 countries is expected to come before the end of this year.
The law began to take shape on April 21, 2021. That’s when the Commission suggested reforming the law. artificial intelligence banning systems like biometric identification or likely to damage social score human rights. This concern focused much of the debate among lawmakers. Therefore, the document signed in June chooses to classify this technology according to the potential impact of its use. There are four categories depending on the type of risk: unacceptable, high, low or minimal.
The regulation would veto any systems that pose “unacceptable risk” and “threat to people”. As the European Parliament proposes, predictive policing or face recognition. The latter will be restricted when in real time (unlike the previous ones) European People’s Party), but this will continue to be allowed in public access areas. More than 155 human rights organizations, including Amnesty International, have called for a complete ban on biometric identification.
Permitted uses
Other uses of AI will also be allowed, but they must be subject to varying degrees of control. The strictest would be for “high risk” systems; security The label covers tools applied in areas such as education, employment or management of critical infrastructures. These need to be assessed, recorded in community data, and met before being commercialized to avoid harm. Transparency criteria The European Parliament includes here: algorithms Content recommendation that will intensify scrutiny of major content platforms Internet those who use them instagram And Google with tiktok And YouTube.
However, the text brings with it many discussions. A coalition of more than a hundred NGOs has condemned that Article 6 introduces a “legal loophole” that allows AI companies to decide whether the systems they develop should be regulated as “high risk”. “If this is approved, the law will become useless,” he explained. Caterina Rodelli, Access Now analyst, to EL PERIÓDICO. They also warn that 27 members of the community club represented by the Council of Europe are putting pressure on negotiations so that the use of artificial intelligence in border control does not have to comply with the same transparency rules.
Generative AI
What will happen? ChatGPT? According to the position adopted by the European Parliament, instruments Generative AI Like the chatbot developed by OpenAI These are in the “limited risk” category. They must only comply with transparency obligations and clearly identify artificially created content. “Users must be aware that they are interacting with a machine in order to make an informed decision whether to continue or step back,” the text reads. The statement is included. The purpose of this measure is to prevent the proliferation of counterfeit products.
According to documents obtained by Time magazine, OpenAI’s push could have helped change an initial draft of the law that classified “general-purpose” AIs like ChatGPT in a “high-risk” category. This would allow him to own this nascent company. MicrosoftAvoid greater restrictions and demands.
The law would also force the companies that create these tools to disclose whether they have been trained with EU-protected material. Copyright. Members of the legislative committee that originally drafted the proposal wanted a complete ban on this use without permission, but ended up requiring transparency.
The debate is in danger
Finally, at the bottom rung will be artificial intelligence that poses “minimal risk” or no risk at all. This category includes the vast majority of systems used in the EU, from automatic spam filters. video game those who use this technology.
“The law aims to stop the negative externalities of AI, but its imminent regulation could also stop beneficial uses,” says technology analyst Antonio Ortiz. ” EU It is in full discussion on how to edit quickly and also how to be. technological powerBecause their position in the development of artificial intelligence systems is not very good.”