Black Forest Labs: Germany's Visual AI Champion and the Scorsese Deal
On 2 June 2026 Black Forest Labs brought in director Martin Scorsese as an adviser, who uses the FLUX image model to storyboard his next film. Behind it stands a German company founded in 2024 and valued at 3.25 billion dollars. This article explains what Black Forest Labs is, what it means for the European AI market and what opportunities and risks visual AI brings for business.
Black Forest Labs from Freiburg is one of the few German AI companies at the technical frontier. On 2 June 2026 it brought in director Martin Scorsese as an adviser, who uses the FLUX image model to storyboard his film with Leonardo DiCaprio. Founded in 2024 by former Stability AI researchers, the company was valued at 3.25 billion dollars in December 2025 after a 300 million dollar Series B round. The FLUX models have been downloaded more than 400 million times and sit inside products from Adobe, Canva, Meta and Microsoft. The case shows both the strength of German AI research and the weakness of the European market: the capital comes from the US, and Europe invests roughly twelve times less in late-stage rounds. For business, visual AI is becoming a practical tool, yet open legal questions around training data and the EU AI Act demand clear governance.
What sits behind the Scorsese deal
A German AI company has moved into the spotlight by way of Hollywood. Black Forest Labs, founded in Freiburg in 2024, brought in director Martin Scorsese as an adviser on 2 June 2026. Scorsese uses generative AI in the form of FLUX to pre-visualize scenes from his next film and to communicate his visual ideas to cast and crew more quickly.
Specifically, Scorsese uses FLUX for storyboarding the drama "What Happens at Night" with Leonardo DiCaprio and Jennifer Lawrence. His stance: "Cinema is a young medium, only around 125 years old, so we have to be open to how it can evolve." BFL co-founder Robin Rombach calls the collaboration a proof point that the technology works in practice.
Cinema is a young medium, only around 125 years old, so we have to be open to how it can evolve.
Martin Scorsese, film directorThe deal is more than a marketing coup. That a director known for his craft brings a German AI model into his workflow shows how far generative image AI has come. It is leaving the playground and becoming a tool in professional production.
Who Black Forest Labs is and what FLUX can do
Black Forest Labs is one of the few European providers working at the technical frontier of generative AI. The company was founded in August 2024 by Robin Rombach, Andreas Blattmann and Patrick Esser, who were previously central to Stable Diffusion at Stability AI. The team is small, with around 70 people, but its reach is large.
The commercial base is broad. FLUX already sits inside products from Adobe, Canva, Meta, Microsoft, ElevenLabs and Vercel, which embed the model into their creative and content workflows. A text-to-video model is in development, and the Scorsese use case points the way. FLUX targets the same niche as Google's Nano Banana or OpenAI's image generators , with the difference that many FLUX models are available as open weights. For how the flagship model works in an enterprise setting, see the hands-on piece on Black Forest Labs FLUX 2 Pro.
German and European perspective
Black Forest Labs shows that a world-class AI provider can emerge from Germany. At the same time the case exposes the weakness of the European market: the growth capital comes almost entirely from the US. Europe invests roughly twelve times less in late-stage rounds than the US.
How differently the path of a German AI champion can run is clear from the contrast with Aleph Alpha. Once hailed as Europe's great AI hope, the Heidelberg firm has been merging with the Canadian provider Cohere since April 2026, after which its existing shareholders hold only around 10 percent. Black Forest Labs remains independent for now.
For Germany as a location, this is an ambivalent story: technical strength meets a structural funding gap. To go deeper, innobu offers an analysis of EU tech sovereignty in chips, cloud and AI and on why France acts while Germany hesitates .
What it means for business
Generative image and video AI is becoming a practical tool for marketing, product development and brand communication. With models like FLUX, product visuals, campaign motifs and prototypes can be created faster and at lower cost. The advantage of open models: they can run in your own or in European cloud infrastructure, which eases data protection and control.
Marketing and sales
Campaign motifs, social media visuals and variant tests take hours instead of days. The brand sets the direction, the model delivers drafts.
Product and e-commerce
Product images, backgrounds and image variants for the online shop can be created without an elaborate photo shoot and adapted to markets.
Concept and pre-visualization
Storyboards, mood boards and concept images make ideas visible early, as the Scorsese use case in film shows.
The role of people stays central. The AI delivers drafts, while the final selection, curation and brand fit remain with the team. The cost advantage over classic production is real, but it does not replace brand understanding. Those planning the step to moving images will find more in the piece on text-to-video with Sora 2 .
Visual AI cuts the cost of image production sharply, but the value only emerges through clear brand and quality rules. Without them, the model produces a lot, but nothing distinctive.
Challenges and risks
Using visual AI raises legal and ethical questions that companies should clarify before deployment. At the centre are copyright and the disclosure of training data. Four points deserve a sober assessment.
Training data is not disclosed
FLUX's training data is not public. That is the central point in the copyright dispute: more than 70 lawsuits are running worldwide against AI providers over the unauthorized use of protected works. Even though Black Forest Labs is not named among them so far, the underlying question concerns the whole industry.
The EU AI Act demands more disclosure
The EU AI Act has obliged providers of general-purpose models since August 2025 to publish a summary of the copyright-protected training content. Anyone using FLUX should know the model and licence documentation and document their own use.
Artists and the industry are divided
The artistic resistance is real. Director Guillermo del Toro firmly rejects generative AI, while others such as James Cameron support it. Companies should design their use so that human creative work is complemented rather than devalued, also to avoid reputational risk.
Provenance and dependence
Without watermarks or provenance data, generated content is hard to tell apart from real material. Standards like C2PA help mark content as AI-generated. And whoever relies on a single model carries provider and licence risk, for example with price changes or revised terms of use.
Do not underestimate reputational risk: AI images without labelling or with questionable provenance can cost trust. A clear internal rule on labelling protects the brand and the customer relationship.
What companies should do now
Companies should evaluate visual AI in a structured way, rather than ignoring it or using it without control. The first step is a clearly scoped test case with measurable value. In parallel, you need rules for rights, labelling and quality control. Five steps help.
-
Set up a pilot
Choose a concrete use case with a clear success metric, such as product images for the online shop or variants for a campaign. That makes the value measurable before you invest more broadly.
-
Check rights and licences
Clarify the model's licence terms and the usage rights to the generated images. Watch for commercial use, redistribution and the question of who owns the results.
-
Mark provenance
Introduce an internal labelling rule for AI content and use provenance standards like C2PA. This creates clarity toward customers and eases later review.
-
Set governance
Define who approves AI images and which content is off limits. A short, binding policy prevents sprawl and protects the brand.
-
Factor in sovereignty
Check whether open models in European infrastructure reduce your dependencies. That keeps you able to act if a provider's prices or terms change.
Anyone aligning use with the rules in force should match it from the start against the deadlines of the EU AI Act for high-risk systems . For a broader market view of the provider race, see the piece on China's AI models in the benchmark test .
Further reading
Frequently asked questions
Black Forest Labs is an AI company founded in Freiburg in 2024 that builds the FLUX family of models for generative image and video AI. Its founders Robin Rombach, Andreas Blattmann and Patrick Esser were previously central to Stable Diffusion at Stability AI. In December 2025 the company was valued at 3.25 billion dollars after a 300 million dollar Series B round.
FLUX is Black Forest Labs' family of models that generate images from text descriptions. The models have been downloaded more than 400 million times and rank among the most widely deployed image generators. FLUX.2 improves in-image text rendering, consistency across multiple reference images and delivers resolutions up to 4 megapixels. A text-to-video model is in development.
Martin Scorsese has been an adviser to Black Forest Labs since 2 June 2026. He uses FLUX for storyboarding and pre-visualizing scenes from his film What Happens at Night, starring Leonardo DiCaprio and Jennifer Lawrence. Scorsese sees the model as a tool to communicate visual ideas to cast and crew more quickly. For Black Forest Labs the deal is proof of the technology's practical maturity.
Yes. Black Forest Labs is based in Freiburg im Breisgau and was founded by German researchers. Its capital, however, comes mostly from the United States, including Andreessen Horowitz, NVIDIA and Salesforce Ventures. That makes the company an example of the strength of German AI research and at the same time of Europe's dependence on US funding.
The main risks are copyright and regulatory. FLUX's training data is not disclosed, while the EU AI Act requires a summary of copyright-protected training content for general-purpose models. Add to that questions of labelling AI content, provenance and dependence on a single provider. Companies should clarify rights, labelling and governance before deployment.