AI Act is one step closer to becoming legislation with EU Parliament approval.

 

A draught of the EU's AI Act has been accepted by the European Parliament, marking a significant advancement towards what may turn out to be the first thorough set of rules for AI in the West.

The final vote count for the measure was 499 in favour, 28 against, and 93 abstentions, therefore it did not pass with a unanimous majority. While today's vote represents a significant advance in the regulation of AI technology, the proposed legislation is still open to amendment because it must first receive approval from all EU member states in order to become a law.

In April, the European Parliament's members initially agreed on its AI Act. after a number of last-minute additions concerning the control of generative AI and huge language models, including ChatGPT. This includes a requirement for generative AI systems to abide by transparency criteria by disclosing if content was AI-generated and assisting in the differentiation between deep-fake and real-world photos.

The complete ban on biometric monitoring in public places and so-called "social scoring" systems, which categorise people based on their social behaviour, socioeconomic level, and personal qualities, are some of the more contentious regulations that are anticipated to face some opposition.

Controversial ban on AI for biometric purposes

The European People's Party faction of parliamentarians has suggested that a complete prohibition on biometric use could impede efforts to combat terrorism and solve crimes.

The law further specifies that any AI systems that could be used to sway voters or the results of elections will be labelled as "high risk," as well as any AI systems that pose a serious risk to people's health, safety, basic rights, or the environment.

"The AI Act will set the tone worldwide in the development and governance of artificial intelligence, ensuring that this technology, set to radically transform our societies through the massive benefits it can offer, evolves and is used in accordance with the European values of democracy, fundamental rights, and the rule of law," said EU co-rapporteur Dragos Tudorache in remarks published along with the announcement by the Parliament.

According to Tim Wright, partner in charge of technology and artificial intelligence legislation at UK law firm Fladgate, organisations that create and use EU systems—including those based outside the EU but providing AI systems and services to EU citizens—will need to pay attention to how the AI Act develops.

Wright compared the rules' implementation to the EU's GDPR and emphasised that businesses should not wait until the last minute to plan, prepare, and implement them because the regulations are intricate and organisations that must comply with them must make sure they can meet the EU's compliance deadline.

According to Wright, "just like with GDPR, non-compliance with the AI Act will come at a significant cost, with failure to meet data governance and high-risk AI transparency obligations attracting penalties of up to €20 million [US422 million] or 4% of global turnover," noting that this is double the fine for failing to meet most other EU requirements.China is the only significant industrialised nation to date to have enacted rules governing generative AI, mandating AI companies to submit security evaluations to the government before disseminating their AI tools to the general public. Additionally, any content produced by generative AI must adhere to the nation's essential socialist principles; failure to do so will result in penalties, service suspensions, or criminal investigations for the providers.