French companies keen to be at the forefront of artificial intelligence — and to rein in European Union regulatory ambitions — have a surprising opponent: the European Commission’s self-styled digital enforcer and über-Frenchman Thierry Breton.
Speaking before an audience of startups and tech investors in Marseille in late November, the EU’s internal market chief called out the French AI hopefuls who are lobbying to erode the EU’s landmark Artificial Intelligence Act.
Breton singled out Mistral — a French AI startup backed by prestigious Silicon Valley investors — putting it in the same category as the U.S. Big Tech firms EU officials usually love to lambast.
“Mistral is lobbying —that’s normal,” Breton said at the event, organized by French newspaper La Tribune. “But we are not fooled: it is defending its business, not the general interest.”
Breton qualified his remarks. He said that Kyutai, a newly launched AI laboratory backed by two French tycoons, Xavier Niel and Rodolphe Saadé, and former Google executive Eric Schmidt is “in the general interest and totally open, it is an initiative that interests and pleases me.”
Breton was stating the obvious, but his words were met with bafflement by France’s AI ecosystem.
“We kind of had the idea that he was on our side,” says the public affairs manager of a French tech company attending the event in Marseille, granted anonymity to discuss the issue candidly. “Obviously, Mistral AI represents private interests — but its work is just as open source as Kyutai’s.”
Many French companies, Mistral chief among them, have over the past few months argued that regulating advanced AI models would deter European innovation. Their pleas found a sympathetic audience in the French government, which alongside those of Germany and Italy is pushing back against rules on “foundation models” proposed by the European Parliament.
The term foundation model is Brussels-speak for the digital infrastructure underpinning mighty multipurpose AI systems such as OpenAI’s ChatGPT. Such models are widely considered the engines of the ongoing AI gold rush.
The Parliament would like to create binding rules of transparency but also of testing on the most powerful models, while governments propose voluntary codes of conduct.
EU policymakers are nearing the end of negotiations on the bloc’s AI rulebook, but the clash over foundation models could still derail a final deal.
Breton’s fighting words are unlikely to earn him many friends in Paris.
President Emmanuel Macron, a tech enthusiast who used to wax lyrical about remaking France into a “startup nation” has thrown his weight squarely behind French AI companies, and has volunteered to host the third AI Safety Summit in November 2024. Last month, speaking in front of researchers, Macron advocated for a global approach to AI regulation to avoid hampering European companies.
Macron’s digital minister Jean-Noël Barrot has repeatedly lambasted the European Parliament’s regulatory plans, illustrating France’s obsession with not missing out on the AI gold rush as it did for the internet and social media. Barrot’s predecessor, Cédric O, sits on Mistral’s board.
There is another reason Breton’s comments are going to rankle. For some time, France’s strategy hinged on luring young AI developers away from U.S. giants. Exhibit A is Mistral itself, whose three co-founders hail from Google and Meta.
But in the past few months, France has grown increasingly willing to lean on U.S. companies to break the technological ceiling and build its ecosystem.
That was on full display recently, during the saga of Sam Altman’s — later reversed — ouster as CEO of OpenAI, when Barrot publicly invited him to move to France.
The shift was partly masterminded by another French golden boy, researcher and Meta’s Chief AI Scientist Yann LeCun. As early as 2015, LeCun played a key role in kickstarting Meta’s Paris research lab, FAIR.
“France is lucky to have him. He is one of the fathers of AI, which made him a tutelary figure in the industry,” says a startup lobbyist granted anonymity as he was not authorized to speak publicly.
That is not to say that Breton is the only one fighting in AI regulation’s corner.
Mark Brakel, policy lead at global AI-focused think tank the Future of Life Institute said that Breton is “completely right” and that lobbying by Mistral and other tech companies “risks sacrificing the benefits of the many for the profits of the few.”
The French cultural sector has also expressed strong concerns that France will sacrifice intellectual property on the altar of technological progress.
“This is the first time that France, where copyright was invented, has not defended intellectual property,” Pascal Rogard, president of the authors’ society SACD, told Le Monde.
Last month, some 80 unions and organizations representing the cultural sector sent a letter to the government to express their opposition to the French stance on the AI Act.
As these concerns grow, the French AI ecosystem is ready to fight back. Just last month, Yann Le Cun had a harsh warning for an event organized by platform regulator Arcom on what would happen if copyright is enforced for generative AI models.
“I’ll tell you right now, the AI industry stops. It can’t work without it,” he said.
Source : POLITICO