Draft rules for artificial intelligence are facing a crucial vote in the European Parliament committee on Thursday, highlighting the urgency for countries to develop regulations for the emerging technology. The proposed rules, known as the EU’s Artificial Intelligence Act, will apply to any product or service that incorporates an artificial intelligence system. The system will be rated on a scale from “minimal” to “unacceptable” danger.Stricter requirements will be imposed on riskier AI systems, such as increased transparency and accurate data use. Currently, most AI systems, including video games or spam filters, fall under the low- or no-risk category.
One of the EU’s primary aims is to prevent AI from threatening human health and safety, while upholding fundamental rights and values. Therefore, the legislation will outright ban dangerous AI applications like “social scoring” systems and interactive toys that encourage harmful behaviour. Predictive policing technology and remote facial recognition will also be prohibited, except in extreme cases like preventing a specific terrorist threat. However, the upcoming vote will determine to what extent the prohibition will apply. This ban aims to prevent the risk of society becoming controlled by AI.
AI systems that fall under high-risk categories such as employment and education will also face stringent requirements like being transparent with users and implementing risk assessment and mitigation measures. Negotiators have added provisions to specifically include general-purpose AI like ChatGPT, and subjecting them to the same requirements as the high-risk systems. One of the crucial additions includes documentations that will allow content creators to determine if their blog posts, digital books, scientific articles or pop songs have been used to train algorithms that power systems like ChatGPT.
The EU’s role in cutting-edge AI development is not as significant as the US and China. However, the EU could still play a global trendsetting role in developing regulations that tend to become de facto global standards. The EU aims to instill confidence in users and develop the market by setting common rules for AI. It is expected that companies and organizations operating within the bloc’s single market with 450 million consumers will choose to comply with EU regulations than having to develop different products for different regions.
Companies like Google and Microsoft could be fined billions of dollars for breaking the AI guidelines, which set the maximum penalty at €30 million ($33 million) or 6% of annual global revenue. It is uncertain when these rules will be fully imposed. However, the legislation is expected to pass by the end of the year or early 2024. Afterward, a grace period is granted for organisations and companies to adapt, which could take up to two years.