{"slug": "project-tapestry", "title": "Project Tapestry", "summary": "The AI Alliance, a non-profit coalition of more than 200 member organizations, launched Project Tapestry, an open-source platform for globally federated development of frontier AI models that enables partners to co-train a shared foundation model while retaining control of their data and building sovereign derivatives. The project aims to achieve frontier performance through consortium-managed data and distributed training, with only model weights shared, and is led by Yann LeCun, Chief Science Advisor.", "body_md": "# Project\n\nTapestry\n\nA consortium approach to training frontier foundation models and sovereign derivatives.\n\nYann LeCun · Chief Science Advisor\n\n## AI Alliance launches Project Tapestry to build a collaborative foundation for open and sovereign AI.\n\nThe AI Alliance, a non-profit AI research and open-source technology coalition with more than 200 member organizations, introduces Project Tapestry — a new open-source platform for globally federated development of frontier AI models — preserving local control and long-term independence.\n\nToday, open-weight AI models are everywhere. But open weights alone do not make pretraining participatory.\n\nBuilt on advances in distributed training which have demonstrated that globally federated model development can match synchronous baselines, the project enables institutions, industries, and nations to co-train a shared open foundation model while retaining control of their data and the ability to build sovereign derivatives aligned to their own priorities.\n\nThe thesis is testable and the science is ready. What remains is coordination.\n\n## The time is now for frontier sovereign AI\n\nToday, models with frontier performance are owned and controlled by a small number of organizations based in a few regions. Only a fraction of the world's data, compute infrastructure, and technical community are involved in frontier model development.\n\nWhat if we could build a new global foundation model with this broader set of knowledge, resources, people and organizations working together — achieving performance no single organization could replicate, that every participating organization could build on?\n\n## Performance is the path to sovereignty\n\nTapestry brings together data, compute and talent from a global consortium to build the next frontier of foundation models — trained on a larger, more diverse corpus than ever before.\n\n### Frontier performance\n\nA global foundation model system more performant than any other, while enabling sovereign derivatives owned and controlled by partners.\n\n### Sovereignty\n\nA person, organization, or nation can own and control their AI, use it how they see fit, and retain the value they create with it — independence and agency.\n\n### Structural advantage\n\nConsortium-managed access to data creates a structural advantage. Distributed ownership, sustainable funding, and mission-locked governance ensure trust.\n\n## Own your data, share in the model\n\nConsortium model\n\n### One base model, many owners\n\nPartners keep full ownership of their data and compute — and the value they create with it. Every contribution makes a base model that all owners build on stronger.\n\nOperating model\n\n### Decentralized contribution, centralized integration, trusted governance\n\nThe design principle behind Tapestry's operating model — contribution happens everywhere, technical integration happens at the core, and governance is trusted by everyone.\n\nDecentralized contribution\n\nCentralized technical integration\n\nGlobally trusted governance\n\nDistributed training\n\n### Only the weights are shared\n\nData can be contributed privately to the core training corpus for base-model training only. Sensitive data is held back at the partner node and used locally to update weights — **only the weights are shared.**\n\nSovereign derivatives\n\n### Your model, fully owned\n\nAny partner can create a derivative model of the base that they fully own and control — using the same infrastructure.\n\n## Globally Sourced Training Data\n\nTraining data sourced from partners around the world may become one of Tapestry's most important assets and the source of our differentiation. As a starting point we seek to catalog potential sources of training data from partners globally.\n\n### What can be contributed\n\n- Public or private datasets\n- Domain / industry data\n- Language & cultural resources\n- Government / public records where lawful\n- Evaluation datasets & environments\n- Preference & feedback data\n\n### What must be described\n\n- Provenance & ownership\n- License & use restrictions\n- Privacy / PII status\n- Quality & AI-readiness\n- Geography & jurisdiction\n- Permitted use (pre/post-training, evals, synthetic gen)\n\n### Benefits to contributors\n\n- Recognition & contribution accounting\n- Access to improved base models\n- Right to create local derivatives\n- Control over restricted data use\n- Path to commercial & sovereign ROI\n- Governance input\n\n**Metadata-first contribution** — partners describe contributions by type and metadata; data stays local and raw corpora are never shipped.\n\n[Propose a dataset →](https://thealliance.ai/projects/tapestry/training-data-proposals?hsLang=en)\n\n## Phase 0: First technical milestones\n\nSpecific early goals to validate hypotheses and de-risk scale-up.\n\n### Cultural alignment\n\nRelease and publish the first model re-aligned to 2+ cultures without core performance loss.\n\n### Distributed training\n\nDemonstrate and open source the first multi-node weight update and aggregation framework; run experiments to understand scale-up requirements.\n\n### Training data catalog\n\nCreate a registry of metadata associated with candidate data sources available to the consortium.\n\n## Long-term roadmap\n\nProgression between phases is gated by technical proof points, partner commitments, compute availability, data readiness, and funding thresholds.\n\n### Initial commitments\n\nMOUs, initial sponsors, tech demos, data catalog prototype; formation steering committee.\n\n● Active now\n### Training platform v1\n\nN-node training, local tuning, initial aggregation, first legal/data framework; initial governance & entity.\n\n### First base model\n\nSmall model from scratch, continued pretraining pipeline, first sovereign derivatives.\n\n### Early deployment\n\nIndustry/government use cases, larger runs, developer & partner ecosystem.\n\n### Frontier-scale effort\n\nContingent on compute, data, and capital thresholds.\n\n## Who should join\n\nTapestry is assembling its first contributors — the researchers, systems engineers, compute providers, governments, and institutions who will build the first sovereign federated training run.\n\n- 01ML researchers working on distributed optimization, federated learning, or low-communication training\n- 02Systems engineers with experience in large-scale GPU cluster orchestration\n- 03Compute providers — cloud, sovereign cloud, or national HPC centers\n- 04Government and policy leaders responsible for national AI strategy\n- 05Universities and research labs with multilingual, domain-specific, or institutional datasets\n\n[Paris workshop official report](https://thealliance.ai/blog/project-tapestry-the-path-to-frontier-sovereign-ai?hsLang=en)\n\n## Join us to build frontier sovereign AI\n\nPath 01\n\n### Learn more / get involved\n\nNew to Tapestry? Tell us how you'd like to participate and we'll be in touch.\n\nPath 02\n\n### Contribute\n\nExplore the code, follow the work, and open issues or pull requests on the Tapestry repository.\n\nPath 03\n\n### Propose data\n\nHave a data source for the consortium? Tell us about it through the form.\n\n### Get involved\n\nTell us how you'd like to participate and we'll be in touch.", "url": "https://wpnews.pro/news/project-tapestry", "canonical_source": "https://thealliance.ai/projects/tapestry", "published_at": "2026-09-08 10:16:55+00:00", "updated_at": "2026-09-08 10:33:59.228030+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "ai-infrastructure", "ai-policy"], "entities": ["AI Alliance", "Project Tapestry", "Yann LeCun"], "alternates": {"html": "https://wpnews.pro/news/project-tapestry", "markdown": "https://wpnews.pro/news/project-tapestry.md", "text": "https://wpnews.pro/news/project-tapestry.txt", "jsonld": "https://wpnews.pro/news/project-tapestry.jsonld"}}