Your RAG’s Secret Backdoor: Leaking Data Through Vector Databases
This article exposes a vulnerability in Retrieval-Augmented Generation (RAG) systems, where misconfigured vector databases can lead to sensitive data leakage. By improperly securing these databases, attackers can gain access to internal documents such as HR policies and top-secret product roadmaps. The RAG system works by storing document chunks as embeddings in a special-purpose vector database and querying it to provide context for the LLM. The focus on securing the LLM while neglecting the vector database leaves it vulnerable to data exfiltration. The attacker can exploit weak access controls and clever retrieval attacks to gain access to sensitive data. Key lesson: Secure vector databases to prevent data breaches caused by RAG system vulnerabilities. #BugBounty #ArtificialIntelligence #DataLeak #Infosec #VectorDatabase

https://infosecwriteups.com/your-rags-secret-backdoor-leaking-data-through-vector-databases-c33755a5abbe?source=rss------bug_bounty-5

[zvec - 초경량·초고속 인프로세스 벡터 DB

Zvec는 초경량·초고속 인프로세스 벡터 DB로, Alibaba의 Proxima 엔진 기반으로 구축되어 대규모 유사도 검색을 최소 설정으로 수행하도록 설계되었습니다. 밀집 및 희소 벡터를 모두 지원하며, 하이브리드 검색 기능을 통해 의미적 유사도와 구조적 필터링을 결합한 정밀 검색을 제공합니다. C++ 기반 핵심 엔진과 SWIG·Python 바인딩 구조로 구성되어 고성능 연산과 다양한 언어 통합을 지원하며, Apache-2.0 라이선스로 제공됩니다.

https://news.hada.io/topic?id=27147

#vectordatabase #algorithms #machinelearning #opensource #search

via @dotnet : Vector Data in .NET – Building Blocks for AI Part 2

https://ift.tt/VtJUvye
#VectorData #NET #AI #BuildingBlocks #SemanticSearch #RAG #Embedding #Embeddings #VectorDatabase #Qdrant #Redis #CosmosDB #SQLServer #PostgreSQL #SQLite #InMemory #VectorSto…

Did you know? Our pgedge-vectorizer tool (on GitHub: https://github.com/pgEdge/pgedge-vectorizer) automatically chunks text content and generates vector embeddings with the help of background workers.

OpenAI, Voyage AI, and Ollama are supported as embedding providers, and a simple SQL interface allows you to enable vectorization on any table. (There’s even built-in views and functions for monitoring queue status.)

#github #opensource #semanticsearch #vector #vectordatabase #openai #ollama #voyageai

[Show GN: OneRAG - 챗봇PM이 만든, 설정 한 줄로 벡터DB/LLM 등이 교체 가능한 RAG 소스

비개발자 크몽 프로젝트에서 시작된 OneRAG는 설정 파일 한 줄만 바꾸면 다양한 RAG 컴포넌트를 교체할 수 있는 오픈소스 도구입니다. Weaviate, Chroma, Pinecone 등 다양한 벡터DB와 Gemini, OpenAI, Claude 등 다양한 LLM을 지원하며, FastAPI 기반으로 Docker Compose를 포함하고 있습니다. 한국어 NLP 기본 지원 및 MIT 라이선스로 제공됩니다.

https://news.hada.io/topic?id=26538

#rag #openai #llm #vectordatabase #chatbot

PostgreSQL with DiskANN indexing now beats Pinecone by 28x on latency at 75% lower cost. AdwaitX analyzes how OpenAI scaled to 800M users and why developers consolidate AI workloads. Technical breakdown #AdwaitX #PostgreSQL #VectorDatabase #AI

https://www.adwaitx.com/postgresql-ai-applications-vector-database/

If you're located near Illinois, Shaun Thomas will be presenting on "The New Postgres AI Ecosystem" at the Illinois Prairie PostgreSQL User Group this February 18th at 5:30 PM CST. 🐘

Come by the DRW and say hi: https://www.meetup.com/illinois-prairie-postgresql-user-group/events/312929674/

#postgresql #postgres #ai #vectordatabase #pgvector #vectorization #aidev #illinois #chicago