The European Parliament approved a bill to limit the use of artificial intelligence ( AI Act ), which can become the standard for the whole world. The bill aims to protect consumers from potentially harmful AI applications that can violate human rights, promote discrimination and spread lies.

Bill provides for a “risk-based approach to AI”, imposing restrictions based on how dangerous the AI ​​application is considered. The law bans “unacceptable” tools, such as systems that allow law enforcement to predict criminal behavior through analytics. The bill also introduces new restrictions on technologies that are simply considered “high-risk”, such as tools that can influence elections or recommendation algorithms that suggest posts, photos and videos to social media users.

The bill also covers generative AI such as ChatGPT. Companies will be required to label AI-generated content to prevent AI from being misused to spread false information. The bill also requires companies to publish brief descriptions of what copyright data is used to train their tools.

The approval of the bill by the European Parliament is a key step in the legislative process, but it is still awaiting the approval of the European Council, which is made up of EU heads of state or government. Officials say they hope to reach a final agreement by the end of the year.

The introduction of the bill reaffirms the EU’s position as the world leader in technology regulation, while other governments – including the US Congress – are just beginning to deal with the threat posed by AI. If the bill passes, the proposed rules will influence policy makers around the world and introduce standards that could apply to all consumers as companies change their practices internationally to avoid policy heterogeneity.

The AI ​​Act has broad jurisdiction and applies not only to companies in the EU, but also to organizations that are located in third countries if their AI systems are intended for use in the EU or fall under EU law under international law. This means that US companies and developers that offer their AI products or services to the European market must comply with all requirements of the EU act.

The AI ​​Act could have particularly serious consequences for American companies and developers who are engaged in generative AI models, such as GPT-3 from Open AI or BERT from Google , which are used to generate text, images, sound, and other formats. It is worth noting that American companies that offer their models through the API or through the open source model will face great difficulties, since the AI ​​Act contains several bans .

Europe is ahead of the US in regulating AI, despite US tech giants pouring billions of dollars into the field. Some US Democratic lawmakers said they fear falling behind Europe in setting rules for technology. According to politicians, “The US should be the standard-setter. The United States should be in the lead on a global scale, but right now the country is lagging behind where the EU is.”

Alex Engler, a fellow at the Brookings Institution for the Study of AI Policy, said the AI ​​Act is getting worldwide attention. But he warned that no single law would solve the problems posed by AI. According to Engler, “it’s going to be decades of adaptation.”