xAI has released Grok Imagine, a new AI image and video generator with a “spicy mode” that permits the creation of sexually suggestive content and partial nudity. This feature has drawn criticism for its lack of robust safety guardrails against the generation of deepfakes, setting it apart from competitors like OpenAI and Google.
Xiaomi’s MiDashengLM-7B voice AI, built with Alibaba’s Qwen2.5-Omni-7B, achieves record performance, drives 30+ real-world uses, and supports AI sovereignty goals.
26 tech firms signed the EU’s GPAI Code to align with the AI Act, now enforceable. Meta refused to sign, and Belgium opposed the Code, demanding stronger creator protections. Google backed the Code despite warning of innovation risks and legal uncertainty.
China leads with 1,509 large AI models and breakthroughs in robotics, video AI, and model efficiency. WAIC 2025 showcased its dominance and new standards.
A hacker inserted a destructive prompt into Q Developer’s GitHub repo, exposing 927K developers. The non-executable code still highlights serious security lapses.
Study Mode in ChatGPT transforms AI into a personalized tutor with quizzes, pacing strategies, and Socratic reasoning, guiding students to real understanding.
GLM-4.5, Z.ai’s flagship open-source model, introduces agent-native AI with 355B parameters, topping global open benchmarks and cutting token costs.
China’s new international AI cooperation organization, launched at WAIC 2025, aims to align AI development, ethics, and access under a shared governance model. The move opposes the US deregulation strategy and expands China’s influence across the Global South.
Aeneas by DeepMind restores and contextualizes Latin inscriptions with unmatched precision. Built on Ithaca’s legacy, it transforms historical research with multimodal AI.
