Adobe launches AI Foundry for Enterprise Brands to Solve AI’s Copyright Risk

• Adobe Launches AI Foundry for Enterprise Model Customization and Commercial Safety.

  Adobe introduced Adobe AI Foundry, a premium model customization service intended to attract more enterprise customers to its ecosystem. The Foundry creates deeply customized, bespoke versions of the flagship generative AI model, Firefly, by re-architecting and retraining the model specific to the client. This service is positioned as a “deep tuning” method, which Hannah Elsakr, vice president of GenAI New Business Ventures at Adobe, explicitly distinguishes from simple fine-tuning. Adobe is positioning this service as a “safer alternative” to mainstream AI tools because its base Firefly models, launched in 2023, were trained entirely on licensed data, and the Foundry models are guaranteed commercially safe, avoiding IP concerns associated with models scraped from the open internet. Adobe will deploy this specialized version through its API solution, Firefly Services.

• Multimodal Capabilities and Brand IP Integration.

  Unlike standard custom Firefly models, which are limited to single-concept image ingestion and response, the AI Foundry versions are multimodal and are designed to understand multiple concepts, allowing them to generate text, images, audio, video, and other mediums like 3D scenes. These models are retrained using the client’s proprietary data, including brand tone, image and video style, products, services, and all Intellectual Property (IP). This deep integration allows the resulting model to generate content for any company use case based on its unique corporate knowledge, enabling highly personalized commerce in an on-brand way.

• Service Addresses Exploding Content Demand and Competitive Differentiation.

  According to a study published by Adobe on Thursday, close to three-quarters (71%) of marketers surveyed said that the demand for content will grow more than fivefold between now and 2027. The AI Foundry allows a customer to create an ad campaign for a product once and use the custom model to help generate the same ad for different seasons, languages, or formats. Enterprises have already used the core Firefly models to create more than 25 billion assets since the 2023 launch, demonstrating high existing demand. The service uses a consumption-based pricing structure that varies by customer and service type, departing from Adobe’s traditional seat-based licenses to offer flexibility.

• Early Adoption by Major Retail and Entertainment Brands, Paired with Complementary AI Tools.

  The service has already secured high-profile early customers, including retail giant Home Depot and Walt Disney Imagineering, the research and development division responsible for Disney’s theme parks. Elsakr likened AI Foundry to an advisory service with teams working directly with clients. Complementary to the Foundry service, Adobe also announced the general release of its LLM Optimizer tool last week (on October 14, 2025), which helps brands track and measure their visibility across AI chatbots like ChatGPT, Gemini, and Claude.

• Data Security and Ownership Protocol for Enterprise IP.

  Adobe has established a strict protocol for handling sensitive enterprise data and Intellectual Property (IP). According to Hannah Elsakr, Adobe will “retrain our own Firefly commercially safe models with the enterprise IP” but will “keep that IP separate” and “never take that back into the base model.” Crucially, the enterprise itself owns that output. The training pipeline involves Adobe’s embedded teams working with the company to identify, securely transfer, ingest, and tag the necessary data before the pre-training model run begins.

Why This Matters:

The launch of Adobe AI Foundry signifies a crucial turning point in enterprise generative AI, moving beyond simple fine-tuning to deep model re-architecture. This service directly addresses the chief concerns of large corporations: brand consistency, IP security, and full ownership of generated assets, particularly by mitigating the significant legal risk associated with training data provenance. By creating highly specialized, multimodal models integrated with a company’s entire IP portfolio and offering a pay-per-use pricing model for scalable production bursts, Adobe positions itself as an indispensable partner for major global businesses, setting a new, higher standard for how AI foundation models must adapt to complex corporate creative and marketing demands while asserting the primacy of human creativity.

This article was drafted with the assistance of generative AI. All facts and details were reviewed and confirmed by an editor prior to publication.