BRUSSELS — A year after the European Union enacted a pioneering strategy to mitigate the risks associated with artificial intelligence, it is now contemplating a significant delay in its implementation. This change comes as calls to pause the rollout of substantial portions of the EU’s Artificial Intelligence Act for 2024 gain momentum, fueled by intense lobbying and rising apprehensions.
The EU’s executive branch has signaled openness to such a pause, and as a final decision approaches, even those against altering the legislation are privately acknowledging the prospect of some form of postponement.
Potential delays and industry reactions
One of the proposals under consideration would allow companies that violate regulations concerning the most high-risk AI applications to maintain their operations for an additional year beyond the original schedule. This shift in stance marks a striking change for Europe, transitioning from a leader in AI regulation to a region wary of lagging behind the U.S. and China in capitalizing on technological advancements.
For many industry leaders, a decision to amend the law would underscore concerns that the EU had not delivered a coherent, enforceable regulatory framework last year that effectively balances risks and business needs. Former Italian Prime Minister Mario Draghi, who recently outlined strategies to enhance the EU’s economic competitiveness, criticized the law as a “source of uncertainty,” further fueling discussions about a delay.
Upcoming deadlines and consultations
A critical date on the horizon is August of next year, when regulations will take effect for AI systems potentially posing “serious risks” to health, safety, and fundamental rights. These include commonly used AI tools in sectors like human resources, education, and the judiciary, which will be required to adhere to various risk management and record-keeping standards prior to market introduction.
However, companies are still waiting for the technical standards necessary for compliance, which have yet to be released. Industry groups and EU member states have emphasized the need for these standards to be finalized well before the August deadline.
In July, leading CEOs in Europe advocated for a two-year moratorium “to address the uncertainty.” Over the past six months, the EU Commission’s perspective regarding a potential pause has evolved. Henna Virkkunen, the EU’s technology chief, indicated in June that if standards aren’t prepared on time, “we should not rule out postponing some parts of the AI Act.” She has since set a deadline for a decision by late August regarding the preparedness of these standards.
As the deadline approaches without a final assessment, urgency has escalated. Draghi publicly called for a pause on high-risk AI regulations “until we better understand the drawbacks” just last week. Additionally, the Commission initiated a consultation to simplify the EU’s tech regulatory framework, suggesting that “targeted adjustments” to the AI Act might be considered as part of this overall strategy.
While the Commission maintains that the upcoming August enforcement date will not be jeopardized, it remains non-committal about specific next steps. European Commission spokesperson Thomas Regnier mentioned, “We have a digital omnibus coming up,” referring to the simplification package expected to be finalized in November.
Poland, which held the Council of the EU presidency until June, has been particularly receptive to delays. It circulated a proposal among other EU countries to postpone penalties for companies that breach high-risk AI regulations by six or twelve months, citing concerns over the law’s “tight implementation deadlines.” The country warned that without adjustments, European startups might be tempted to relocate to “less regulated jurisdictions.” Sweden and the Czech Republic have also expressed support for a pause.
As EU nations prepare to discuss Poland’s proposal, lawmakers in the European Parliament are also gearing up for deliberation. The Parliament’s leading voices on AI have scheduled a meeting for October 15 to examine the initiative to simplify the EU’s tech regulatory framework and its implications for the AI Act. Dutch Greens lawmaker Kim van Sparrentak has voiced strong opposition to any pause, describing it as a “bizarre proposal” at a time when the AI Act has yet to be implemented. Nevertheless, she acknowledged that minimizing potential damage is now a primary concern.
“One of the things that we’re increasingly discussing with the people that are focused on the implementation of the AI Act is: how do we ensure that the damage remains limited?”