Harnessing the Universal Geometry of Embeddings

Researchers introduced the first method for translating text embeddings between vector spaces without paired data, encoders, or predefined matches, achieving high cosine similarity across model pairs with different architectures, parameter counts, and training datasets. The method, described in a paper submitted on 18 May 2025 and revised on 26 Jan 2026, raises security concerns for vector databases, as an adversary with access only to embedding vectors could extract sensitive information about underlying documents.

Computer Science Machine Learning Submitted on 18 May 2025 v1 https://arxiv.org/abs/2505.12540v1 , last revised 26 Jan 2026 this version, v4 Title:Harnessing the Universal Geometry of Embeddings View PDF /pdf/2505.12540 HTML experimental https://arxiv.org/html/2505.12540v4 Abstract:We introduce the first method for translating text embeddings from one vector space to another without any paired data, encoders, or predefined sets of matches. Our unsupervised approach translates any embedding to and from a universal latent representation i.e., a universal semantic structure conjectured by the Platonic Representation Hypothesis . Our translations achieve high cosine similarity across model pairs with different architectures, parameter counts, and training datasets. The ability to translate unknown embeddings into a different space while preserving their geometry has serious implications for the security of vector databases. An adversary with access only to embedding vectors can extract sensitive information about the underlying documents, sufficient for classification and attribute inference. Submission history From: Rishi Jha view email /show-email/465a8825/2505.12540 Sun, 18 May 2025 20:37:07 UTC 3,179 KB \ v1\ /abs/2505.12540v1 Tue, 20 May 2025 15:38:41 UTC 3,180 KB \ v2\ /abs/2505.12540v2 Wed, 25 Jun 2025 21:04:02 UTC 2,407 KB \ v3\ /abs/2505.12540v3 v4 Mon, 26 Jan 2026 14:47:13 UTC 2,424 KB References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender IArxiv Recommender What is IArxiv? https://iarxiv.org/about arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .