The European Parliament's Civil Liberties (LIBE) Committee and Internal Market (IMCO) Committee overwhelmingly supported a bill that would establish a risk-based framework for regulating the application of artificial intelligence. In a vote this morning, the commission voted 71 to 8 (with seven abstentions) in favor of the compromise plan negotiated with EU member states in a long-running trilateral “tripartite talks” late last year.
The EU AI Law was originally proposed by the European Commission in April 2021 and would set rules for AI developers based on the capabilities of their models and the purposes for which they are applying AI. This includes a list of prohibited uses of AI (e.g., social scoring) and a defined set of high-risk uses (e.g., education, health, Contains regulations regarding employment, employment, etc. Additionally, it applies some transparency requirements to general purpose AI and tools such as deepfakes and AI chatbots.
Additionally, most AI apps are considered “low risk,” meaning they are outside the scope of the law. The plan also provides for the establishment of national regulatory sandboxes to allow developers to develop, train and test dangerous apps in supervised “real-world” environments.
The European Commission's proposal for an AI rulebook did not cause much uproar when the EU proposed it three years ago. But with the rise of generative AI over the past year+, the plan has gained global attention and caused a major rift among lawmakers in the bloc. MEPs have moved to amend the proposal to ensure it applies to strong general AI, but a small number of member states, led by France, are hoping to develop national champions. are moving in the opposite direction, calling for separate regulations for advanced AI. .
At the marathon tripartite talks in December, a compromise proposal was submitted that still included provisions regarding general-purpose AI, resulting in prolonged opposition from some governments. And as of late last month, it appeared those disagreements could still prevent the bill from passing. But a significant vote by member states on a compromise passed earlier this month makes it almost certain that they will adopt the flagship AI rulebook within months.
That said, the bill still faces some hurdles before it can be adopted. A plenary vote will be held in Parliament in the coming weeks, at which MPs will be asked to formally adopt the bill. Final board approval will then take place.
However, these last few measures seem unlikely to lead to an upset among the EU's co-legislators. Such a move would be a wrecking ball for the bill in the current cycle, as Congressional elections loom and the university's current mandate ends. This means that legislative time and reputational room for adjustment are both tight.
Today's strong support from two parliamentary committees that have been involved in detailed consideration of this bill for years also sends a strong signal that MPs will stick to the required absolute majority. This will pave the way for the bill to be adopted and come into force later this year. The first provision (prohibition of prohibited conduct) will apply six months later (probably later this year).
The EU has agreed to a phased rollout of the law, with legal requirements likely to be imposed on targeted developers between 2024 and 2027. ()The EU also allows the Code of Conduct to apply for nine months after it enters into force. General purpose AI rules, including governance, will apply 12 months after they take effect. It also states that the law will become fully applicable 24 months after it takes effect. However, the applicable period for the obligation for high-risk systems is longer (36 months). )
Although there is a high possibility that the bill will be passed by the plenary session of the parliament, there is strong opposition among some. The Pirate Party, for example, refuses to endorse the members who hold the most votes on the committee today. against This law has been called a “defective” law. Pirate Party lawmaker Marcel Colaya, who is also a member of the IMCO committee, said in a statement: Therefore, the Pirates cannot support it. ”
Pirate Party MP and LIBE member Patrick Breyer also warned in a separate statement: “EU AI law opens the door to permanent facial surveillance in real time. She is wanted on more than 6,000 European arrest warrants for crimes listed in the AI Act. For these reasons, every public space in Europe could be placed under permanent biometric mass surveillance. This law justifies and normalizes a culture of mistrust. It leads Europe into a dystopian future of a high-tech surveillance state filled with mistrust. ”