Europeans Push Regulatory Threat in First-of-Its-Kind AI Act

The European Union is one step closer to regulating artificial intelligence with a first-of-its-kind AI Act which sets out to ban any AI applications that pose a risk to the public at large. The act places AI in one of three categories identifying it as an unacceptable risk, a high-risk application, or largely left unregulated.

The AI Act site clarifies that under the first unacceptable risk category, it will ban applications that put the government at risk, such as “government-run social scoring of the type used in China.” Meanwhile, if an AI application is placed in the high-risk category, the act specifies that these applications, like CV-scanning tools, will go under review to ensure they meet legal requirements.

The act includes a wide range of demands and restrictions on AI products including increased transparency requirements for notifying users about the source of generated information and greater scrutiny over applications that would impact official government functions. 

“We’re making history!” Parliament President Roberta Metsola said in a Twitter post on Wednesday. She continued, “Proud that @Europarl_EN has just approved the world’s first Artificial Intelligence Act. A balanced and human-centered legislation which will set the global standard for years to come.”

Metsola and the EU’s Parliament did not immediately respond to Gizmodo’s request for comment.

The AI Act still requires one final review by the EU Parliament’s executive branch in what’s called the “trilogue” stage of review. The act will begin its review with officials on Wednesday night and once lawmakers vote on it, the AI Act will proceed to the European Parliament, European Commission, and the Council of the European Union who will have until January to approve it.

The EU has been increasingly cautious when it comes to AI, voicing concerns that it could violate the Personal Data Protection law that prohibits developers from launching applications that could share an individual’s personal data. Applications including Microsoft’s ChatGPT and Google’s Bard chatbot have faced increasing difficulties rolling out their software in the EU which required the companies to present documentation proving it complied with local regulations.

If passed, the AI Act will streamline the process, giving the EU a more straightforward approach to banning or regulating AI applications. Yet, for all its good intentions, the AI Act does face a series of loopholes, one of which presents a major concern that there is no room for leeway once passed. The AI Act site says, “If in two years’ time, a dangerous AI application is used in an unforeseen sector, the law provides no mechanism to label it as ‘high-risk.’”

Despite the law’s shortcomings, parliament is optimistic that the AI Act “could become a global standard” around the world. “This moment is hugely significant,” Daniel Leufer, a senior policy analyst focused on AI at Access Now’s Brussels office told Time. “What the European Union says poses an unacceptable risk to human rights will be taken as a blueprint around the world.”