Text Embeddings Reveal (Almost) As Much As Text. Embeddings of text—where a text string is converted into a fixed-number length array of floating point numbers—are demonstrably reversible: “a multi-step method that iteratively corrects and re-embeds text is able to recover 92% of 32-token text inputs exactly”.
This means that if you’re using a vector database for embeddings of private data you need to treat those embedding vectors with the same level of protection as the original text.
Recent articles
- Nano Banana Pro aka gemini-3-pro-image-preview is the best available image generation model - 20th November 2025
- How I automate my Substack newsletter with content from my blog - 19th November 2025
- Trying out Gemini 3 Pro with audio transcription and a new pelican benchmark - 18th November 2025