The EU’s upcoming AI law could be “a zero or one regulation” for European generative AI startups like Mistral, said Cédric O, the former French government minister who’s been leading the Paris-based company’s public affairs efforts.
“Depending on the [final shape of the] AI Act, it could kill Mistral, it could allow Mistral to grow, and there is a lot of gray zone in between,” O, who used to be Macron’s minister for digital affairs and is now a cofounder at the buzzy AI startup, tells Sifted in an interview. “It's really critical.”
Mistral, which is building large language models, is considered to be one of the few European tech companies that could rival American AI giants such as OpenAI, Google and Meta. But the six-month-old startup, which recently raised a gigantic €105m seed round, is concerned that the outcome of ongoing negotiations around the EU’s flagship AI regulation might pull the carpet from under its feet.
After four years of intense negotiations, the EU is expected to finalise its rules for AI companies by the end of this year. The new regulation, the first of its kind in the world, is likely to put extra requirements, such as reporting on data and algorithms, on companies that use AI classified as “high risk” and also, in some cases, on companies that produce powerful AI foundation models — such as Mistral.
This approach has been frowned upon by many European startups which often don’t have the resources to deal with additional bureaucratic burdens. Research conducted by a group of European AI associations at the end of 2022 found that 50% of startups surveyed thought the AI Act would slow down AI innovation in Europe, while 16% said they were considering stopping developing AI or relocating outside the EU.
O says that the new wave of AI innovation is an opportunity for European companies to emerge as leaders on the global technology scene — and it’d be a shame to lose this chance because of strict regulations.
“The main problem for Europe is not regulation: the main problem for Europe is that in the digital world, the leaders set the standard, and Europe has no leaders,” he says, pointing to the world’s largest tech companies, which are either American or Chinese. “Europe should be focusing on this one thing: how to make European champions emerge.”
But O believes that, with the new crop of European AI startups, the continent might finally have a chance.
“I've been working to see potential European champions being able to compete on a par with American competitors and we are seeing that generation emerging. Will they win? We don't know. But at least for the first time over the past 20 years, we can play the game,” he says.
“Basically, what I think is: don't shut off the light now — and that was the message I've been trying to convey over the past six months on the European scene.”
Regulate it well
O adds that he isn’t saying that there shouldn’t be any regulation at all, but that tech sovereignty should be taken into consideration.
“We're not asking for special rules, because we're Europeans trying to catch up with Americans. We’re just asking to have the same empowerment as the Americans, the same freedom as the Americans,” he adds. “If we have the same freedom, then we're going to be able to compete… We're just asking to have the same leeway as the others.”
He adds that Mistral, similar to other European startups, doesn’t have the resources to fight the EU's bureaucratic and legal machine, especially in comparison to the thousands of lawyers at the disposal of the big tech companies in the US and Asia.
“We’re 20 [people]. We’re ambitious, but we’re still 20,” he says. “If the burden that is on the shoulder of the foundation model provider, both from the bureaucratic or legal liability point of view, is too heavy, then it could kill Mistral.”