Google removed its previous policy prohibiting the use of artificial intelligence for weapons purposes in February 2025, coinciding with the release of its annual Responsible AI report. The company's new AI principles are built on three main pillars: bold innovation, responsible development and deployment, and progress based on
The development of large language models has accelerated in recent months. The HUN-REN Hungarian Research Centre for Linguistics has now introduced the PULI LlumiX model, which is a system optimised for the Hungarian language, continuously pre-trained and fine-tuned for instruction following. A detailed presentation of the model can be found
The first international AI safety report was published on 29th January 2025 by 96 international experts led by Yoshua Bengio, documenting artificial intelligence's social, economic, and environmental impacts with concrete data. The document jointly endorsed at the 2023 Bletchley Park AI Safety Summit provides a scientific foundation for
Beyond their underlying architectures, another crucial way to classify generative systems is by the type of data—or modality—they process and generate. Previously, such models were typically limited to a single modality (unimodal), meaning they were designed to process and generate only one type of content. A prominent example
According to OpenAI experts, the Grok 3 artificial intelligence performance data published by xAI may be misleading, as they have questioned the credibility of the published test results, particularly regarding scores achieved on the AIME 2025 mathematics test. The dispute centres on the fact that the graphs published in xAI&
Sam Altman, CEO of OpenAI, announced the roadmap for new artificial intelligence models on February 12, 2025: GPT-4.5 will arrive within weeks, while GPT-5 will be available months later, in the second half of 2025. The announcement is particularly significant as OpenAI is radically simplifying its product portfolio and
Elon Musk's xAI company introduced its latest artificial intelligence model, Grok 3, on February 17, 2025, which the company claims surpasses leading AI chatbots, including ChatGPT and DeepSeek. The model was developed on the Memphis "Colossus" supercomputer, which utilizes 200,000 Nvidia GPUs, providing ten times
The European Commission established the EU Artificial Intelligence Office (AI Office) in January 2024, which began operations with a team of 140 experts. The office plays a key role in implementing the EU AI Act - the world's first comprehensive AI regulatory framework - with a special focus
Generative models represent a fundamental paradigm in machine learning, enabling computers to create new data samples that closely mirror real-world examples. These models have become indispensable tools across diverse fields including image creation, natural language processing, and scientific research. Three principal architectures have emerged as dominant approaches: Generative Adversarial Networks
The AI Builders Summit 2025 event series showcased the latest directions in artificial intelligence development over four weeks, from large language models (LLMs) to AI agents. During the virtual event organized by ODSC between January 15 and February 6, 2025, hundreds of participants tracked the evolution of various AI technologies,
Anthropic launched the Anthropic Economic Index initiative on February 10, 2025, which examines the economic impacts of artificial intelligence based on millions of actual usage data. The analysis processed one million conversations with Claude and found that AI use affects 36% of various occupations while serving augmentation purposes (57%) more
Researchers from Stanford University, the University of Washington, and the Allen AI Institute have developed a new method to increase artificial intelligence efficiency. The s1 model, which was created with less than $50 worth of computing resources, achieves performance that was previously only possible in projects with significant budgets. The
