Google improves AI-powered weather forecasts
Google’s WeatherNext 3 improves weather forecasting by combining current raw satellite data with global weather analyses. This allows the model to detect atmospheric changes more quickly and update forecasts hourly instead of every six hours as before.
Anthropic links AI agents to the real world
Anthropic has unveiled the Model Hardware Standard (MHS). The new technology is designed to enable AI agents to control physical devices safely. To this end, MHS provides a unified interface between AI models and laboratory or industrial equipment.
Could “Banned Topics” help stop AI hackers?
Researchers suggest that restricting certain topics in AI systems could help prevent cyberattacks. The method uses an approach similar to the techniques hackers use to manipulate AI models, but applies it as a defensive measure.
Europe’s opportunities for AI catch-up
Europe is falling behind the US and China in artificial intelligence. The main reasons are a lack of investment, limited venture capital, and challenges in turning research results into successful companies.
ArXiv tightens AI rules
Research platform arXiv will suspend authors for one year if submitted papers contain clear signs of unchecked AI-generated content. The policy targets researchers who rely too heavily on AI without verifying facts, claims, or citations.
SAP is building a European AI center of excellence
SAP plans to acquire the German start-up Prior Labs and invest around $1.16 billion to build a leading European AI research center focused on enterprise data. Prior Labs was founded in Freiburg just 18 months ago and specializes in “tabular foundation models.”
China is rapidly closing the AI gap
The latest Stanford AI Index Report shows that China has significantly narrowed the gap with the United States in artificial intelligence. In key performance indicators, the two countries are already nearly on par.
Europe leads in AI research, but lags behind in commercialization
Europe achieves outstanding results in research across artificial intelligence, drones, robotics, quantum technologies, and cybersecurity, establishing itself as one of the world’s leading scientific regions.
AI hits limits in certain games
Recent research shows that while modern AI can excel at many games, it often struggles with tasks that require understanding underlying mathematical functions. In games like Nim, current AI training approaches fail.
Mistral releases Mistral 3
French startup Mistral AI released Mistral 3, a comprehensive suite of new open models. The lineup features three „Ministral 3“ models (with 3B, 8B, and 14B parameters) tailored for on-device use, alongside the massive 41B-parameter „Mistral Large 3,“ which utilizes a sparse mixture-of-experts architecture totaling 675B parameters.