Daily summary · 1 November 2026
The most relevant AI news of the last 24 hours. Each one links to its story and to the original source.
- Local & Open Source
Open WebUI v0.12.5 Released with Enhanced Note and Analytics Features
Open WebUI v0.12.5 has been released, adding analytics for administrators to view chat details, a multi-page note system, and live presence features in notes.
Source: Open WebUI - Agents & Dev
MCP Servers v2.2.0 Released: Support for CA Certificate Updates
Following reports that private keys for eMudhra intermediate CA certificates could be recovered, MCP Servers v2.2.0 has been released. It details the responses from Node.js and Python's certifi.
Source: MCP Servers - Research
FDSML: Proposing Formula-Driven Supervised Metric Learning
FDSML is proposed to enable vision pre-training without real images, combining PA-Triplet Loss based on generation parameters with the refined FineFractalDB dataset.
Source: Hugging Face Papers - Agents & Dev
BLA, an adversarial defense shield against AI reverse-engineering, appears on GitHub
On October 25, 2026, a new open-source repository called qybaihe/bla was released, designed to protect binaries, Python, and other assets from AI reverse-engineering agents.
Source: GitHub New AI Repos - Workflows & Nodes
New ComfyUI-based video character swap repository 'openjutsu' released
Released on 2026-10-26, openjutsu is an open-source repository that enables one-click character swapping in videos using ComfyUI, MiniMax H3, and FLUX.3 Klein relighting.
Source: GitHub New AI Repos - Research
Research on Proving and Improving the Platonic Representation Hypothesis with JEPAs
A study has been released that uses JEPAs to prove a Platonic Representation Theorem regarding the hypothesis of whether models trained on different modalities converge to the same representation.
Source: Hugging Face Papers - AI Companies
Databricks and SAP partner to provide data and AI capabilities to accelerate enterprise AI transformation
Databricks and SAP announced they are providing end-to-end data and AI capabilities to accelerate AI transformation for enterprises.
Source: Databricks - Local & Open Source
Ollama v0.40.13 Released with Support for Liquid AI's LFM3
Ollama v0.40.13 has been released, enabling Liquid AI's LFM3 models to run on Linux and Windows. This includes models with text-to-speech capabilities, such as lfm3-audio and lfm3-omni.
Source: Ollama
AI通信について / About / Acerca de · 確認方法 / How we verify / Verificación · 訂正 / Corrections / Correcciones · 応援 / Support / Apoyar · メール / Newsletter / Boletín · プライバシー / Privacy / Privacidad