The European Parliament's civil liberties (LIBE) and internal market (IMCO) committees have overwhelmingly backed a bill establishing a risk-based framework to regulate artificial intelligence applications. In a vote, the committees voted 71-8 (with 7 abstentions) in favor of the compromise negotiated with EU member states late last year in lengthy trialogue talks.
The EU AI Law, originally proposed by the Commission in April 2021, sets out rules for AI developers based on the power of their models and/or the purpose for which they intend to apply the AI. It includes a list of prohibited uses of AI (such as social scoring), along with rules for a defined set of high-risk uses (for example, education, health or employment), as well as obligations in areas such as data quality, testing and risk assessment. Additionally, it applies some transparency requirements to general-purpose AI and tools such as deepfakes and AI chatbots.
Beyond that, most AI applications will be considered “low risk,” meaning they fall outside the reach of the law. The plan also provides for the establishment of nationwide regulatory sandboxes to allow developers to develop, train and test potentially risky applications in a supervised “real-world” environment.
The Commission's proposal for an AI regulation did not cause much of a stir when the EU presented it three years ago. But with the rise of generative AI over the past year, the plan has captured global attention and opened wide gaps between the bloc's lawmakers. MEPs decided to amend the proposal to ensure it applies to powerful general-purpose AIs, while a handful of Member States, led by France, have been pushing in the direction of the opposition, seeking regulation for advanced AIs in the hope to foster national champions. .
Marathon tripartite talks in December resulted in a compromise text that still included some provisions for general-purpose AI, prompting persistent opposition from some governments. And still last month it seemed that these divisions could still derail the passage of the bill of law. But with a critical vote by Member States on the compromise text approved earlier this month, it seems almost certain that the bloc will adopt its flagship AI regulation within a few months.
That said, the bill still has some hurdles to clear before adoption: there will be a plenary vote in parliament in the coming weeks, where MEPs will be asked to formally adopt it. Then there will be a final endorsement from the Council.
However, these latest steps seem less likely to cause disagreements among EU co-legislators. Any such move would be a wrecking ball for the bill in the current cycle, with parliamentary elections around the corner and the end of the term of the current faculty, meaning that both legislative time and reputational margin of maneuver are reduced.
The broad support from the two parliamentary committees, which were involved in the detailed examination of the legislative proposal for years, also gives a strong signal that MEPs will go ahead with the support of the necessary absolute majority, which would pave the way for that the law be approved. and will come into force at the end of this year. The first provisions (bans on prohibited practices) would apply six months later (most likely in the second half of this year).
The EU decided on a phased implementation of the Act that will likely see legal requirements increase for in-scope developers between 2024 and 2027. (The EU also allows nine months after entry into force for a code of practice to apply; 12 months after entry into force for general-purpose AI rules, including governance, to apply. It also says the law will be fully applicable 24 months after entry into force, although obligations for high-risk systems They still have a longer application period (36 months).
While a plenary vote in parliament appears likely to pass, some opposition remains. The Pirate Party, for example, refuses to support what its MEPs (who represent a couple of votes in today's committee) against the law, they describe it as a “defective” law. In a statement, Pirate Party MEP Marcel Kolaja, also a member of the IMCO committee, said: “Unfortunately, despite the good position of the European Parliament, national governments managed to paralyze the AI Law. Therefore, we cannot support it.”
In another statement, Patrick Breyer, Pirate Party MEP and member of the LIBE committee, also warned: "The EU AI Law opens the door to permanent real-time facial surveillance: more than 6.000 people are wanted by European arrest warrant for the crimes listed in the AI Law. Any public space in Europe may be under permanent mass biometric surveillance for these reasons. This law legitimizes and normalizes a culture of distrust. “It leads Europe into a dystopian future of a distrustful, high-tech surveillance state.”
