Spawning

There is a wave of new AI datasets being collected that are free of copyright concerns. Congratulations to Spawning and Cullen Miller for this strong new addition!

We’ve put 12.4 million of the most beautiful images in the public domain into an image-text dataset with community governance. Public Domain 12M is completely free. Its 30+ terabytes of images are hosted on AWS, so downloading them is fast and won’t negatively impact the cultural heritage organizations who originally shared them. Visit https://source.plus/pd12m to explore the dataset, download it through Hugging Face, or read the paper to get into the details. We’ve enabled metadata, semantic, and faceted search so it’s easy to view the entire contents of PD12M. We welcome you to explore and use flagging features responsibly to help us continue to improve the dataset.

Adobe is launching its Content Authenticity Initiative web app in beta in the first quarter of 2025, allowing creators to apply content credentials to their work, certifying it as their own. Content credentials take provenance a step further than altering an image's metadata. Adobe’s system uses digital fingerprinting, invisible watermarking, and cryptographically signed metadata to more securely protect an artwork, including images, video, and audio files. Adobe can “truly say that wherever an image, or a video, or an audio file goes, on anywhere on the web or on a mobile device, the content credential will always be attached to it,” according to Adobe’s Senior Director of Content Authenticity Andy Parsons. Another tool helping artists retain control over how their works are used online is Spawning. Artists can add their works to a Do Not Train registry on its website called “Have I Been Trained?”, which signals to AI companies that this work shouldn’t be included in training datasets. Only effective if AI companies honor the list, but so far, Hugging Face and Stability AI are on board. On Tuesday, Adobe is launching the beta version of the Content Authenticity Google Chrome extension. Creators can also sign up to be notified when the beta for the full web app launches next year. TechCrunch Amanda Silberling October 8, 2024 Coalition for Content Provenance and Authenticity (C2PA) #Art #FineArt #AI #AIArt #ArtFraud #Provenance #Watermark #Adobe #ArtificialIntelligence #NoAI

I am very excited to share the artwork I exhibited in the recent show at Ringling College of Art and Design, "Artificial Visions: Faculty Explorations in AI Artistry". My digital illustration was created in a multi-step process that ensured it was both in my own unique style and intentionally sourced from CC0/Public Domain collections through https://source.plus/ . You can read more about it in my artist's statement below. “Bazaar Before the Storm” My work explores the fusion of traditional and digital generation techniques, blending the tactile intimacy of hand-drawn illustrations with the precision and depth of technology. In creating this image, I begin with the personal—a collection of my own drawings, photographs, and experiences. These elements form the foundation of my work, grounding the digital in the deeply human. My process involves first honing and fine-tuning AI models from my own illustrations and image datasets gathered from the public domain through Source.Plus. These fine-tuned models act as ingredients in a complex recipe of style and content generation. I run multiple custom models simultaneously with an individually controllable degree of influence on the final outputs. Once I have models in place, I use my own photography and digital drawings to create compositional guides and depth maps that allow me to intentionally shape the generated outputs. Landscapes start as simple black and white sketches. Characters are generated directly from a photo of a person or from pose information extracted from the same photo. There is a certain amount of unpredictability to AI image generation and I value those unusual results. Each figure is regenerated many times and I select the best output for the larger composition. Every character and object require slight tweaks to achieve a consistent stylistic result, and I embrace the happy accidents I discover in this technical trial-and-error approach. I find it fascinating to allow the technology to interpret, transform, and elevate my original creations. The AI acts as both a collaborator and a tool, reimagining my drawings and photographs in ways that are both unexpected and profoundly resonant. This process invites a dialogue between the organic and the algorithmic, where the unpredictability of AI meets the intentionality of the artist. Surrealism is at the core of all my artistic exploration. The surrealists were famous for their many “games” to remove conscious control of their art-making processes; to embrace their subconscious mind. I consider AI a sort of external subconscious. ?The juxtaposition of my hand-drawn imagery, public domain illustrations, and machine generated forms creates a sense of otherworldliness; a glimpse into realms that exist beyond the confines of organic imagination. Through this synthesis, I seek to create work that not only reflects my inner world but also challenges the boundaries of what is considered authentic art in the age of artificial intelligence.

Breakfast. Lazy Saturday morning. Late January. "The artists that want to get the power back from the AI companies", read the title in my favourite newspaper, Morgenbladet - I was hooked. The story was about Berlin based artists taking ownership of what they publish online, real solution for opting out from being trained on by #ML algorythms ( as European law requires) and how can you create art together with machines. Who owns the art then? That was relevant for #AI, #creatives, #tech AND #IP nerds! Fast forward several months: Berlin based tech nerd Cullen Miller at Spawning (featured in the article) - check? AI nerd and writer Silvija Seres - check? IP nerd Benita S. Mackay - check? Researcher and AI collaborator Benedikte Wallace - check? Join me during Oslo Innovation Week and hear all of these fantastic people at Forskningsparken - Oslo Science Park. Patentstyret - Norwegian Industrial Property Office (NIPO) is hosting. And giving out free breakfast:)

Spawning recently launched a new project, Source.Plus: “Our goal is to make it easy for rights holders to offer their media for use in generative AI training — on their own terms — and frictionless for developers to incorporate that media into their training workflows," shared CEO and Co-Founder Jordan Meyer. Learn more in TechCrunch from Kyle Wiggers: https://lnkd.in/eUS3U73W #ethicalAI #AItraining

Can't wait to hear more about these fantastic data empowerment projects & tools showcased by Mozilla Festival this Thursday! From trailblazer Spawning's ' 'haveibeentrained.com' & their upcoming 'Source.plus' platform (to curate non-infringing collections for AI training), to the 'Data provenance initiative', the 'FLAIR' initiative by Mila - Quebec Artificial Intelligence Institute ('building the foundations for Indigenous Voice AI in systems explicitly designed to respect data sovereignty and linguistic self-determination') & 'The computational privacy group' from Imperial College London , you don't want to miss this!

We’re excited to extend early access to Spawning’s next project,?Source.Plus?— a platform to curate, enrich, and download non-infringing media collections in bulk for AI training. Source.Plus collections are seeded with nearly 40m high quality public domain and CC0 images, compliant with EU regulations, and more than enough to train a SoTA base model. Our roadmap includes an extended suite of data enrichment and compliance tools for model trainers across modalities, and methods to license premium collections from rights holders. Learn more about it here: https://lnkd.in/ew6NBvn6

One Size Does Not Fit All: Machine-Readable Rights Reservations A look at the benefits and limitations of key machine-readable methods for registering rights reservations https://lnkd.in/eYTiHKzm

Opting out of AI training can be difficult. We built three tools to make it easier. Learn more about how to add media to the Do Not Train Registry in our new post: https://lnkd.in/gAVCRhYR