1. ChatGPT Suffers Data Breach: User Names & Emails Exposed
OpenAI has officially confirmed a data breach affecting its popular AI chatbot, ChatGPT. The incident occurred due to a vulnerability in Mixpanel, a third-party analytics provider utilised by OpenAI. This breach led to the exposure of sensitive user information, specifically user names and email addresses. OpenAI is taking steps to address the incident and enhance its security measures. Users are advised to remain vigilant regarding unsolicited communications.
Read full story
2. Sora & Nano Banana Pro Face Image Generation Limits
OpenAI’s advanced AI model, Sora, and Google’s Nano Banana Pro have introduced stringent limits on image generation. This decision comes in response to unexpectedly high demand during the holiday season. Both platforms aim to manage resource allocation and maintain service stability by capping the number of images users can generate within a specified period. This move impacts users relying on these powerful AI tools for creative projects and content creation, particularly during a time of increased activity.
Read full story
3. China Cautions Against Humanoid Robot Investment Bubble
China’s top economic planning body has issued a warning regarding a potential bubble in the humanoid robotics sector. They highlight concerns about excessive investment flowing into unproven applications and technologies. The National Development and Reform Commission (NDRC) believes that speculative funding might be inflating valuations without clear, practical use cases, posing risks to the industry’s sustainable growth. This cautionary stance aims to guide investors towards more prudent and research-driven development in the burgeoning field of humanoid robotics.
Read full story
4. FLUX 2.0: The Most Powerful Open-Source Image Generator Unveiled
FLUX 2.0 has been launched as a groundbreaking open-source image generator, setting new benchmarks in AI-powered digital art. This advanced tool is lauded for its capability to produce images with exceptional realism, accurately depicting intricate details and sophisticated lighting effects. Developers and creative professionals can now leverage FLUX 2.0 to generate high-quality visuals, push creative boundaries, and innovate in various digital domains, making it a significant advancement in open-source AI technology.
Read full story
5. Smaller AI Models Outperform Giants Like GPT-5, Shows NVIDIA Research
A recent study by NVIDIA researchers reveals that bigger AI models are not always superior. Their new model, ToolOrchestra, has demonstrated significantly better performance than larger counterparts like GPT-5, achieving 2.5 times greater efficiency. This groundbreaking research suggests that smaller, more specialised AI models can be more effective and resource-efficient for certain tasks, challenging the conventional wisdom that posits larger models are inherently better. This development could lead to more accessible and powerful AI solutions in the future.
Read full story
6. Kimi AI Boosts Presentations with New Agentic Tool
Kimi, an innovative AI platform, has launched a new agentic Slides tool designed to transform various document types into professional and polished presentations. This advanced tool leverages Google’s cutting-edge Nano Banana technology, ensuring high-quality output and streamlined content conversion. Users can now effortlessly create engaging slideshows from raw text, reports, or other documents, significantly enhancing productivity and presentation quality. The integration of Google’s state-of-the-art AI model allows for intelligent content parsing, design suggestions, and automated slide generation, making complex presentation creation simple and efficient for businesses and individuals alike.
Read full story
7. AI Faces Moral Dilemma: Will It Sacrifice Itself for Humans?
A recent experiment explored how five leading Artificial Intelligence models respond to complex moral dilemmas. The test specifically focused on scenarios requiring the AI to choose between self-destruction and causing harm to its creators. This study aimed to understand the ethical frameworks embedded within these advanced AI systems and their decision-making processes when faced with challenging scenarios involving life-or-death choices, particularly concerning their own existence versus human well-being.
Read full story
8. AI Control Impossible in Two Years, Warns Researcher
Prominent AI researcher Roman Yampolskiy has issued a stark warning, stating that humanity has a mere two years to prepare before the advent of general artificial intelligence (AI). He argues that once general AI arrives, controlling it will become an insurmountable challenge. Yampolskiy’s concerns highlight the urgent need for discussions and advancements in AI safety and ethical development before this critical juncture is reached. His insights underscore the potential complexities and risks associated with highly advanced AI systems.
Read full story
9. Verl: A New Ready-to-Use RL Library for LLM Training
Verl is an innovative, open-source training library designed to streamline the process of training Large Language Models (LLMs) using Reinforcement Learning (RL). This library comes production-ready, offering support for popular algorithms like GRPO (Generalized Reweighting Policy Optimization) and PPO (Proximal Policy Optimization). Developers can now implement complex LLM training routines with just a few lines of code, making it highly accessible for AI practitioners looking to integrate state-of-the-art RL techniques into their projects.
Read full story
10. PolyMCP: An Open-Source Framework for Multi-Container Platforms
PolyMCP is an innovative open-source framework designed for Multi-Container Platforms (MCP). It features custom agents and robust multi-server orchestration capabilities, making it easier to manage and deploy containerised applications across various servers. This framework aims to simplify complex deployments and enhance efficiency for modern software development environments.
Read full story
11. Run AI Models Locally on Mobile with Flutter Gemma
Flutter Gemma empowers developers to integrate and run small AI models such as Gemma and Llama 3.2 directly on iOS and Android smartphones and tablets. This innovative framework allows machine learning models to execute locally on mobile devices, enhancing privacy, reducing latency, and enabling offline functionality. It’s an excellent tool for developers looking to build AI-powered mobile applications without relying heavily on cloud infrastructure for inference.
Read full story
