The European Parliament took a significant stride towards EU-wide regulation of artificial intelligence (AI) systems such as ChatGPT on Thursday, an initiative that Brussels is eager to implement swiftly. Committees on civil liberties and consumer protection gave a resounding vote for a position text that proposes limits on AI usage in Europe while simultaneously promoting innovation in this sector. The document is set to be presented to the entire parliament next month for approval before proceeding to negotiations with EU member states to draft the final law.
A Historic Vote for AI Rules
Lawmakers have dubbed Thursday’s vote “historic,” hoping it will pave the way for the “world’s first rules on artificial intelligence.” The text incorporates primary elements from a proposal by the European Commission two years prior, suggesting additional prohibitions on biometric surveillance, emotion recognition, and predictive policing AI systems.
A Call for Transparency in AI
The parliament’s proposal aims to categorize generative AI systems such as ChatGPT and Midjourney under a group that requires unique transparency measures. This includes notifying users that the generated content was produced by a machine, not a human.
Defining High-Risk AI
The proposal also outlines extra criteria to determine what constitutes a “high-risk” AI application, potentially narrowing down the scope of this designation. The European Commission’s proposed list includes AI applications in critical infrastructure, education, human resources, public order, and migration management.
However, Members of the European Parliament (MEPs) argue for an additional threshold requiring safety, health, or fundamental rights threats to also be considered.
Industry Reactions to the Proposed Regulations
While the Computer & Communications Industry Association (CCIA) acknowledged some useful improvements in the parliament’s text, it expressed concern over the abandonment of the risk-based structure from the European Commission’s proposal. Boniface de Champris, CCIA policy manager for Europe, stated, “The best way for the EU to inspire other jurisdictions is by ensuring that new regulation will enable, rather than inhibit, the development of useful AI applications.”
On the other hand, the European Consumer Organization welcomed the Parliament’s move towards stronger consumer protection compared to the original Commission’s proposal. Deputy Director Ursula Pachl emphasized the need to protect consumers against potential risks posed by these novel technologies.
AI’s Emergence in Public Consciousness
The discussion on AI regulation has been ongoing, but the actual and potential impact of AI has only come to the forefront of public awareness in recent months, especially with ChatGPT’s emergence last year. AIs like Midjourney and DALL-E, capable of generating impressive images, and AI music sites producing human-like singing, have fueled the conversation.
There is growing concern among policymakers in Europe and globally about the potential misuse of AI technology, particularly in spreading misinformation and influencing public opinion and elections. This worry has prompted calls for a pause until appropriate legal frameworks can be put in place.
Ensuring Compliance with Regulations
The commission’s proposal calls for AI companies to keep human control over their algorithms, provide technical documentation, and manage risks for “high-risk” applications. Each EU member state would appoint a supervisory authority to ensure compliance.
MEPs also propose that AI companies implement measures against illegal content and copyright infringement, particularly in materials used to train their algorithms. They also advocate for prohibiting the scraping of online photos for algorithm training without obtaining the concerned individuals’ consent.