Artificial intelligence isn’t just capable of translating between existing languages—it can also create entirely new ones.Artificial intelligence isn’t just capable of translating between existing languages—it can also create entirely new ones.[#item_full_content]

The era of building “personalized AI” by training AI models on individual or corporate documents and data is beginning. However, while such customization can improve task performance, it can also weaken a model’s existing safety safeguards. KAIST researchers have developed a core AI technology that preserves customized performance while further strengthening safety.The era of building “personalized AI” by training AI models on individual or corporate documents and data is beginning. However, while such customization can improve task performance, it can also weaken a model’s existing safety safeguards. KAIST researchers have developed a core AI technology that preserves customized performance while further strengthening safety.[#item_full_content]

When can we trust the results we get from AI, and when is learning impossible? Researchers have shown that there are some problems that even the most powerful AI cannot reliably solve, no matter how much data it is given.When can we trust the results we get from AI, and when is learning impossible? Researchers have shown that there are some problems that even the most powerful AI cannot reliably solve, no matter how much data it is given.[#item_full_content]

Quantum computers lack useful functionality without the right algorithms to facilitate their operation. Currently, there are few simple, standardized operations, known as “primitives,” in the quantum toolkit that can help deliver the unique quantum behaviors required for such computers to achieve results beyond classical systems.Quantum computers lack useful functionality without the right algorithms to facilitate their operation. Currently, there are few simple, standardized operations, known as “primitives,” in the quantum toolkit that can help deliver the unique quantum behaviors required for such computers to achieve results beyond classical systems.[#item_full_content]

Neural networks, a fascinating technology inspired by the human brain, form the basis of artificial intelligence. These networks consist of layers of interconnected nodes, or artificial neurons, that learn patterns from data and make predictions. For example, large language models generate text by predicting the next word or phrase based on the words that came before it.Neural networks, a fascinating technology inspired by the human brain, form the basis of artificial intelligence. These networks consist of layers of interconnected nodes, or artificial neurons, that learn patterns from data and make predictions. For example, large language models generate text by predicting the next word or phrase based on the words that came before it.[#item_full_content]

Korean researchers have developed a hierarchical AI technology that autonomously plans even complex, long-horizon tasks. The development of this hierarchical task-planning AI technology, which reduces hallucinations and doubles the success rate, is expected to help robots and agents carry out long-term missions.Korean researchers have developed a hierarchical AI technology that autonomously plans even complex, long-horizon tasks. The development of this hierarchical task-planning AI technology, which reduces hallucinations and doubles the success rate, is expected to help robots and agents carry out long-term missions.[#item_full_content]

Training artificial intelligence to enforce even seemingly straightforward rules—like balls and strikes in Major League Baseball (MLB)—is a messy, dynamic process that takes time and careful evaluation of the technology in the wild, according to new Cornell research.Training artificial intelligence to enforce even seemingly straightforward rules—like balls and strikes in Major League Baseball (MLB)—is a messy, dynamic process that takes time and careful evaluation of the technology in the wild, according to new Cornell research.[#item_full_content]

Semiconductors are central to modern technology. They are used in computer chips, solar cells, sensors, LEDs and communication devices. Before researchers make new semiconductor materials in the lab, they often test them first using quantum mechanical simulations. One of the main tools for this work is density functional theory, or DFT, a computer-modeling method that skips tracking every single electron and instead uses their overall cloud density to quickly calculate a material’s atomic structure and energy.Semiconductors are central to modern technology. They are used in computer chips, solar cells, sensors, LEDs and communication devices. Before researchers make new semiconductor materials in the lab, they often test them first using quantum mechanical simulations. One of the main tools for this work is density functional theory, or DFT, a computer-modeling method that skips tracking every single electron and instead uses their overall cloud density to quickly calculate a material’s atomic structure and energy.[#item_full_content]

A study led by researchers from IMDEA Networks and Carlos III University of Madrid (UC3M) has carried out the first large-scale analysis of the volatility, content and actual infrastructure of hidden websites on the so-called dark web. This work was among the recipients of the 2025–2026 Spanish Police Foundation Research Awards, which recognize scientific research that contributes to ensuring public safety and protecting citizens’ rights and freedoms.A study led by researchers from IMDEA Networks and Carlos III University of Madrid (UC3M) has carried out the first large-scale analysis of the volatility, content and actual infrastructure of hidden websites on the so-called dark web. This work was among the recipients of the 2025–2026 Spanish Police Foundation Research Awards, which recognize scientific research that contributes to ensuring public safety and protecting citizens’ rights and freedoms.[#item_full_content]

When a standard large language model (LLM) is confronted with a problem, it tries to solve it by matching it to similar information it has seen before, and then give an answer based on those past patterns. But how it decides which information to use and what value it gives to different pieces of information can be somewhat inscrutable from the outside. An EPFL team has created a new large language model that is structured similarly to a human brain, allowing users more control and moving away from “black box” AI.When a standard large language model (LLM) is confronted with a problem, it tries to solve it by matching it to similar information it has seen before, and then give an answer based on those past patterns. But how it decides which information to use and what value it gives to different pieces of information can be somewhat inscrutable from the outside. An EPFL team has created a new large language model that is structured similarly to a human brain, allowing users more control and moving away from “black box” AI.[#item_full_content]