TL;DR:
- Brussels and US Congress at odds over AI policy, highlighting global regulation challenges.
- EU has led in proposing AI regulation since 2021, facing uncertainty due to lobbying and opposition.
- US Congress is working on bipartisan AI legislation but lacks specifics.
- Key discussions on AI risks and national security were held in Congress.
- Senators aim for an “incentive-based” approach to retain AI developers in the US.
- EU seeks a “risk-based” approach to limit the use of AI applications.
- Tensions arise within the EU over binding regulations and transparency rules for foundation models.
- France is a major obstacle to the EU AI Act due to its interest in AI companies and AI in policing.
- If there is no agreement by January, passing AI legislation before the European Parliament elections could be challenging.
Main AI News:
In the realm of global artificial intelligence regulation, policymakers in Brussels find themselves entrenched in late-night negotiations, fervently seeking to establish a groundbreaking law to govern artificial intelligence. In contrast, across the Atlantic, US Senators are signaling a distinct approach from their EU counterparts, expressing concerns about the potential heavy-handedness of the EU’s approach, which they fear could alienate AI developers.
This dichotomy, unfolding on both sides of the Atlantic, vividly underscores the formidable challenges associated with regulating artificial intelligence—an issue that has gained paramount importance for governments worldwide over the past year, driven in part by the global fascination ignited by AI-powered chatbot ChatGPT.
Brussels has been ahead of the curve since 2021, having initially proposed a framework for AI regulation. However, the fate of the EU’s AI legislation remains uncertain, mired in a formidable lobbying effort and opposition from some of the EU’s major nations, including France, Germany, and Italy.
After months of dedicated work on AI policy, Senate Majority Leader Charles E. Schumer (D-N.Y.) has revealed that a bipartisan group is making significant strides in crafting legislation. However, details on the specifics of such legislation remain scant.
These developments have unfolded during Congress’s final two AI forums of 2023, where lawmakers engaged in discussions with top tech executives, including former Google CEO Eric Schmidt, to delve into crucial topics such as the risks associated with AI doomsday scenarios and national security concerns.
Sen. Mike Rounds (R-S.D.), a member of the bipartisan working group convened by Schumer to formulate AI policy, has revealed that the Senators are pursuing an “incentive-based” approach to retain AI developers within the United States. The objective is to prevent AI development from gravitating towards other nations that may pose potential threats.
Meanwhile, within the European Union, officials are endeavoring to achieve a breakthrough in the late stages of the EU AI Act. This legislation predominantly follows a “risk-based” approach, seeking to restrict the use of AI applications based on the perceived level of danger they pose.
Representatives from the European Parliament are racing to counter efforts by some of the EU’s larger member states to dilute this historic bill. Recent negotiations among various EU bodies—the European Commission, responsible for proposing laws, and the European Council and European Parliament, tasked with adopting them—have been marred by divisions that imperil a bill that has been years in the making. Optimism initially pervaded the negotiations, but as the clock inches closer to midnight in Brussels, talks continue with no guarantee of an agreement.
If the marathon negotiations fail to yield an agreement and stretch into the early hours of Thursday morning in Brussels, the process may be relegated to a last-ditch effort in January. Experts warn that passing any bill before the European Parliament elections in June could become exceedingly challenging.
Brando Benifei, one of the leading lawmakers championing the EU AI Act in the European Parliament, has voiced concerns, stating, “If we go beyond January, I think we are lost. It will be at least another nine months before we could have the AI Act.”
Key nations within the EU are advocating the removal of provisions from the bill that would impose binding regulations and transparency rules on foundation models, such as the technology underpinning ChatGPT. They argue that these rules could stifle innovation and put Europe at a disadvantage compared to the United States in the race to develop such models, favoring industry self-regulation instead.
Sources familiar with the negotiations, speaking on condition of anonymity, suggest that France is the most significant obstacle to reaching an agreement. This is partly due to France’s desire to protect burgeoning companies involved in AI foundation models, including Paris-based Mistral, as well as other French AI firms. Additionally, France’s plans to deploy AI-powered smart cameras for policing and security at the 2024 Summer Olympics, along with the deployment of AI technology in French cities, have further complicated matters.
Addressing French opposition, France’s digital minister, Jean-Noël Barrot, emphasized that European governments widely oppose restrictions on AI’s use for policing and national security. He argued that burdensome regulations on foundational model developers could hinder European innovation.
Barrot asserted that even the compromise sought by France would result in the world’s most robust AI regulation. He viewed the bill as the beginning of European regulations on AI technology, rather than the culmination.
As negotiations continued on Wednesday, Benifei stressed that allowing the industry to self-regulate, as advocated by France and other countries, would negate one of the most vital aspects of the bill. He insisted that imposing real restrictions on the security, transparency, and data usage of powerful AI models would enhance safety throughout the AI ecosystem.
Conclusion:
The divergence in AI policy between the US Congress and the European Union reflects the intricate challenges of global AI regulation. While the EU leads in proposing comprehensive regulation, internal divisions and opposition from key nations pose significant hurdles. In the US, lawmakers are navigating the development of bipartisan AI legislation with a focus on retaining AI talent. The outcome of these regulatory efforts will have a profound impact on the AI market, influencing innovation, development, and competitiveness on both sides of the Atlantic.
