Adobe has firmly committed to not using user content to train its AI models. The company updated its terms of use to clarify that its generative AI tools and models have never been trained on user content and will not be in the future.
The new pledge in Adobe's terms of use aims to legally reassure users. "Your content is yours and will never be used to train any generative AI tool," Adobe stated. "We will make it clear in the license grant section that any license granted to Adobe to operate its services will not supersede your ownership rights."
Adobe's suite of generative AI-powered creative tools includes the image generation model Firefly, which was trained on licensed content with permission and public domain images.
In contrast, other companies like Google and OpenAI use user data to enhance their AI services. Meta has faced criticism and legal challenges in the EU over its AI training practices.
While Adobe uses user data and content characteristics to enhance product features like masking and background removal, it assures that none of this data is used for training generative AI. Users also have the option to opt out of this data usage.
Adobe's commitment builds on a January 2023 promise to avoid using user data for AI training. Scott Belsky, Adobe's chief product officer, emphasized the company's stance: "If we ever allow people to opt-in for generative AI specifically, we need to call it out and explain how we’re using it."
In related news, companies like Nvidia, Adobe, and IBM have agreed to let third parties vet their AI models to ensure transparency and trust in their AI practices.
