Toggle light / dark theme

Get the latest international news and world events from around the world.

Log in for authorized contributors

Language shapes visual processing in both human brains and AI models, study finds

Neuroscientists have been trying to understand how the brain processes visual information for over a century. The development of computational models inspired by the brain’s layered organization, also known as deep neural networks (DNNs), have recently opened new exciting possibilities for research in this area.

By comparing how DNNs and the human brain process information, researchers at Peking University, Beijing Normal University and other institutes in China have shed new light on the underpinnings of visual processing. Their paper, published in Nature Human Behavior, suggests that language actively shapes how both the brain and multi-modal DNNs process visual information.

Synchronizing ultrashort X-ray pulses for attosecond precision

Scientists at the Paul Scherrer Institute PSI have, for the first time, demonstrated a technique that synchronizes ultrashort X-ray pulses at the X-ray free-electron laser SwissFEL. This achievement opens new possibilities for observing ultrafast atomic and molecular processes with attosecond precision.

Scrutinizing fast atomic and molecular processes in action requires bright and short X-ray pulses—a task in which free-electron lasers such as SwissFEL excel. However, within these X-ray pulses the light is internally disordered: its temporal structure is randomly distributed and varies from shot to shot. This limits the accuracy of certain experiments.

To tame this inherent randomness, a team of PSI researchers has succeeded in implementing a technique known as mode-locking to generate trains of pulses that are coherent in time. “We can now obtain fully ordered pulses in time and frequency in a very controlled manner,” says accelerator physicist Eduard Prat, who led the study, published in Physical Review Letters.

Going further with fusion, together

At 4 a.m., while most of New Jersey slept, a Princeton Plasma Physics Laboratory (PPPL) physicist sat at his computer connected to a control room 3,500 miles away in Oxford, England. Years of experience running fusion experiments in the U.S. helped guide the U.K. team through delicate adjustments as they worked together to coax particles of plasma—the fourth state of matter—to temperatures that match those found at the heart of the sun.

This late-night, intercontinental collaboration happened many times from 2019 to 2024 during critical experiments at Tokamak Energy’s ST40 facility. It’s just one example of how PPPL is meeting the moment, leading collaborative efforts with private companies and other public institutions to make fusion power practical.

Fusion, the process of combining atoms to release energy, could be the source of a nearly inexhaustible supply of electricity. But there are still challenging scientific and engineering issues to overcome in the quest for power. That’s why scientists are increasingly working together to take fusion further.

3D-Printed “Light Cages” Could Solve One of Quantum Networking’s Biggest Problems

A new chip-based quantum memory uses 3D-printed “light cages” to store light in atomic vapor with high precision. Quantum information storage plays a central role in the development of the quantum internet and future quantum computers. Today’s quantum communication systems are limited by signal l

Hubble’s Newest Discovery Isn’t a Star, It’s a Window Into the Dark Universe

Scientists have identified a strange cosmic relic called Cloud-9 — a starless, gas-filled object dominated by dark matter.

Detected with Hubble, it appears to be a failed galaxy that never formed stars, preserving a snapshot of the early Universe.

A starless relic revealed by hubble.

Black Cat Behind SEO Poisoning Malware Campaign Targeting Popular Software Searches

The malware establishes contact with a hard-coded remote server (“sbido[.]com:2869”), allowing it to steal web browser data, log keystrokes, extract clipboard contents, and other valuable information from the compromised host.

CNCERT/CC and ThreatBook noted that the Black Cat cybercrime syndicate has compromised about 277,800 hosts across China between December 7 and 20, 2025, with the highest daily number of compromised machines within the country scaling a high of 62,167.

To mitigate the risk, users are advised to refrain from clicking on links from unknown sources and stick to trusted sources for downloading software.

New GoBruteforcer attack wave targets crypto, blockchain projects

A new wave of GoBruteforcer botnet malware attacks is targeting databases of cryptocurrency and blockchain projects on exposed servers believed to be configured using AI-generated examples.

GoBrutforcer is also known as GoBrut. It is a Golang-based botnet that typically targets exposed FTP, MySQL, PostgreSQL, and phpMyAdmin services.

The malware often relies on compromised Linux servers to scan random public IPs and carry out brute-force login attacks.

Critical jsPDF flaw lets hackers steal secrets via generated PDFs

The jsPDF library for generating PDF documents in JavaScript applications is vulnerable to a critical vulnerability that allows an attacker to steal sensitive data from the local filesystem by including it in generated files.

The flaw is a local file inclusion and path traversal that allows passing unsanitized paths to the file loading mechanism (loadFile) in jsPDF versions before 4.0. It is tracked as CVE-2025–68428 and received a severity score of 9.2.

The jsPDF library is a widely adopted package with more than 3.5 million weekly downloads in the npm registry.

/* */