‘Everybody is just kind of holding their breath’ What to watch for the EU AI Act’s key meeting this week
All eyes in European tech will be on Brussels Wednesday — in particular, those building companies in the hot AI space. The reason? It’s crunch time for the much-debated EU AI Act.
To catch you up: Europe’s first attempt to regulate artificial intelligence has faced a lot of pushback recently from countries like Germany and France over last-minute additions around regulating the development of large language models, or LLMs — the tech behind popular chatbots like OpenAI’s ChatGPT — versus just regulating applications of the technology. The proposed rules could be a big problem for startups like Germany’s Aleph Alpha and France's Mistral, both of which have raked in huge sums of cash this year to develop their models. If the proposed changes pass, startups like them face additional layers of bureaucracy.
When policymakers meet on December 6 for an end-of-year debate on the Act, they could finalise it — or the sticking points could push the process way into 2024.
“Everybody is just kind of holding their breath a little bit”, Jeannette zu Fürstenberg, founding partner of Berlin-based VC firm La Famiglia (which recently merged with General Catalyst), tells me. She’s been involved in the pushback against the latest draft, and although La Famiglia is an investor in Mistral, Fürstenberg says the resistance is about more than making sure the company survives from an investment point of view. It’s about “securing future European competitiveness” in the AI race, she says.
Fürstenberg argues that regulating a foundation model is like regulating a coding language — “just because you can do things with it doesn't mean that you're responsible for what can be done with it,” she argues. Regulation is needed for applications of the technology, she concedes, adding that she believes founders should also be thinking through the negative side effects of the ways they’re applying the technology.
But not everyone agrees that the models themselves should be given free rein. "Having seen the potential systemic risks that those systems could introduce, it makes sense that we demand regulation,” Carme Artigas, Spain’s AI minister — who has been involved in the negotiations on behalf of the Council of the EU — told Sifted last month. Others argue that if LLMs aren’t regulated now, other startups that rely on the technology (like those using OpenAI’s GPT-4 for their businesses) might be stuck with the legal ramifications.
If policymakers don’t land on a finalised set of principles and rules on December 6, the Act could be delayed until after the election of a new EU parliament and commission next year.
Fürstenberg says it’s very uncertain. Others aren’t optimistic.
“An agreement by the end of the year currently seems rather unrealistic,” Christoph Stresing, managing director at Startup-Verband, Germany’s startup association, recently told Sifted. “In our view, there must be considerable improvements to the AI Act in the further negotiations [...] It is better to do it well than to rush it. Because the AI Act is too important for bad compromises.”
Certainly startups — and especially founders with millions in their coffers earmarked to develop their AI models — will be eagerly watching what happens this week. And, of course, so will I. If you’re a founder or VC with thoughts on how the EU AI Act might impact the ecosystem — for good, or ill — get in touch.
— Anne Sraders, senior reporter
.png?upscale=true&width=600&upscale=true&name=Sponsor%20Card%20-%20Flagship%20AMPLY%20(1).png)
