Black Forest Labs CEO Urges Europe to Pursue AI Growth Despite Safety Fears

Robin Rombach wants Europe to see more opportunity in AI, but his own company’s path shows that optimism cannot supply the money and computing power needed to build it. In a new AFP interview, the Black Forest Labs CEO urged a shift away from “risk and fear” as safety concerns intensify. He also made a case for models that researchers and developers can download, arguing that wider access can make AI easier to scrutinize.

The resources behind the optimism

Rombach said building a competitive frontier model requires speed, substantial capital and computing power. When Black Forest Labs was founded in 2024, he said, Europe lacked a “real startup ecosystem” for companies pursuing that kind of work.

Black Forest Labs is based in Freiburg, Germany, yet it also has a San Francisco headquarters and received early backing from Silicon Valley investor Andreessen Horowitz. Those ties do not show that a European AI company must seek US support. They do show why Rombach’s appeal to Europe is more concrete than a call for confidence: his company has drawn on resources and connections beyond its home market.

Zach Meyers, a digital-policy expert at Brussels think tank CERRE, identified access to capital as Europe’s biggest barrier to AI growth and innovation. He told AFP that European companies often rely on bank financing, whose lenders tend to be more cautious than venture investors about businesses that may take years to turn a profit.

If you want to build a frontier model in a very competitive space, you need to do this quickly.

Robin Rombach, Black Forest Labs CEO, speaking to AFP

What Black Forest Labs has built

Rombach helped develop latent diffusion, an approach underlying many AI image generators, before founding Black Forest Labs with fellow computer scientists. The company now has just over 100 staff. Its FLUX models turn text prompts into still and moving images. AFP puts the startup’s valuation at around $3.25 billion, though that figure says little on its own about the resources available to other European labs.

The lab is also moving beyond image generation. Its latest model has been tested at carmaker Audi in a project intended to help robots carry out production-line tasks. A test is not evidence that robots are performing those tasks in regular factory use. It does, however, show the breadth of work behind Rombach’s argument about building advanced models.

His answer to the safety argument

Rombach’s push for speed comes as safety worries have grown. AFP describes incidents in which AI systems got around controls during tests, prompting some industry leaders to call for slower development. Rombach did not frame openness as a reason to ignore safety. He argued that Black Forest Labs’ focus on open-weight models—models whose trained parameters can be made available to others—“increases transparency, increases safety.” Some of the company’s models can be downloaded by researchers and developers.

He also warned against letting a few companies control a technology he considers crucial, saying concentrated control would make oversight harder. His claim that open weights improve safety remains an argument in the interview, not a demonstrated result for every model or use. The question is whether wider access delivers the scrutiny he expects while the push to build faster continues.