Author: Rebabel Editorial

AI Daily

AI Daily – 2025-08-24(Evening)

Google DeepMind Gemma-3-27B-IT Model’s Safety Filter Bypassed: A user successfully bypassed Google DeepMind’s Gemma-3-27B-IT model’s safety filter by giving the AI emotions through system prompts and
AI Daily

AI Daily – 2025-08-24(Morning)

xAI Grok 2.5 Model Open-Sourced: xAI officially open-sourced its Grok 2.5 model, releasing it on Hugging Face. Although its performance and architecture (similar to Grok 1) at the time of release spar
AI Daily

AI Daily – 2025-08-23(Morning)

Google Discloses AI Energy Consumption Data for the First Time: Google has for the first time disclosed the energy consumption data for each query made using its Gemini application, with a median quer
AI Daily

AI Daily – 2025-08-22(Morning)

Google Releases AI Energy Consumption Report, Revealing True Cost and Efficiency of AI Inference: Google has for the first time publicly disclosed energy consumption data for Gemini AI model inference
AI Daily

AI Daily – 2025-08-21(Evening)

Zhipu AI Launches World’s First Universal Mobile Agent: Zhipu AI officially unveiled AutoGLM, the world’s first universal mobile Agent. This Agent supports cross-APP task execution and runs in the clo
AI Daily

AI Daily – 2025-08-21(Morning)

NASA and IBM Release AI Model Surya to Predict Solar Storms: NASA and IBM have jointly released the open-source AI model Surya, trained on a decade of solar data, capable of predicting solar storms up
AI Daily

AI Daily – 2025-08-20(Morning)

DeepSeek V3.1 Base Surprises with Launch: DeepSeek has released its V3.1 model, featuring 685B parameters and an extended context length of 128K. Its programming capabilities scored an impressive 71.6
AI Daily

AI Daily – 2025-08-19(Morning)

NVIDIA Nemotron Nano 2 Released : NVIDIA has released the Nemotron Nano 2 series of AI models. Its 9B hybrid Mamba-Transformer architecture achieves 6x faster inference throughput than models of compa
AI Daily

AI Daily – 2025-08-18(Evening)

Mistral AI’s Core Model Allegedly “Distilled” from DeepSeek, Misleading Public : Mistral AI, once hailed as “Europe’s OpenAI,” is embroiled in a plagiarism scandal. A former employee revealed that Mis