Welcome to our latest round-up of news from the technology and hosting world. Here’s what we’ve discovered this month.
AI Ransomware
Cybersecurity company ESET has discovered the first-ever ransomware developed using open-source AI tools. Named PromptLock, the malware uses OpenAI’s newly launched gpt-oss:20b model to create Lua scripts in real time. These scripts can scan the local filesystem, extract chosen files, and perform encryption across Windows, Linux and macOS systems.
Developed using the Go programming language, PromptLock uses the SPECK encryption algorithm to lock victims’ files. Ironically, SPECK itself was developed by the US National Security Agency (NSA).
While PromptLock has not yet been detected in a cyberattack, its appearance indicates a major advancement in ransomware development. Where AI has previously been used by cybercriminals to improve phishing and social engineering, PromptLock integrates AI directly into its operational framework.
The discovery of PromptLock has raised concerns that attackers will start to utilise private or local AI systems to bypass security and automate code generation when carrying out ransomware attacks. Experts warn businesses to strengthen security in preparation for more AI-driven threats.
Hydrogen-Powered USV
Plymouth-based startup, Acua Ocean, has successfully tested its hydrogen-powered unmanned surface vessel (USV), the Pioneer. Monitored remotely from the company’s headquarters, the test saw the boat complete a 24-hour mission, operating autonomously 12 miles off the UK coast. The mission was accomplished with zero emissions.
In what is a world-first, the project has been approved by the UK Maritime and Coastguard Agency, with the Pioneer being certified under the Workboat Code. The regulatory approval is a major step forward in the wider use of hydrogen propulsion in commercial shipping and the reduction of emissions in maritime operations.
While the project is still in its early stages, the successful deployment of the Pioneer shows the potential that remote, autonomous, clean-powered vessels have in shipping, offshore support and environmental monitoring.
UK Police Robot
Nottinghamshire Police is testing a robotic dog to see how it can help in high-risk policing situations. Equipped with multiple cameras, a LiDAR scanner and a loudspeaker, the four-legged robot can send live video, detect weapons and let officers talk to suspects from a distance. Able to operate over a range of more than a kilometre, the robot can also climb stairs, navigate obstacles and fit into tight spaces that are inaccessible to drones.
Over the three-month trial, Nottinghamshire Police’s firearms training team will test how the robot can be deployed in events like sieges, hostage situations and chemical or biological threats. Additionally, the team will look at how it can keep officers safe through scouting, carrying supplies and monitoring areas remotely.
While similar law enforcement robots have been trialled in other countries, this is the first time they have been tested by UK police. If found to be effective and affordable, the robots could potentially be deployed across the country as soon as next year.
Exascale Supercomputer
The Jupiter supercomputer, housed at the Jülich Supercomputing Centre in Germany, has succeeded in exceeding one quintillion operations per second, making it Europe’s first exascale supercomputer and the fourth fastest in the world.
Built on Nvidia’s Grace Hopper platform and Eviden’s BullSequana XH3000 liquid-cooled architecture, Jupiter features 24,000 Grace Hopper Superchips and over 50,000 high-speed network connections utilising Quantum-2 InfiniBand. Another key feature is its energy efficiency. It can achieve more than 60 billion floating-point operations per watt, and it employs warm water cooling to recycle waste heat into the local heating system.
A €500 million investment from Germany and the EU will see Jupiter conduct advanced research in climate science, medicine and particle physics. It will also be used to support the creation of sophisticated European language models. The system is expected to boost AI innovation and strengthen Europe’s digital independence, offering researchers and industries access to sustainable, large-scale computational resources.
Next-Gen GPUs
Nvidia has revealed its next-generation Rubin GPU series, specifically designed to manage large-context AI tasks, like advanced coding and high-speed video creation. The new processors will operate alongside the Vera CPU within the Vera Rubin NVL 144 CPX platform, which offers 8 exaflops of AI computing power – more than seven times the performance of the company’s current GB300 NVL72 systems. With each rack providing 100TB of rapid memory and a bandwidth of 1.7 petabytes per second, it sets a new standard for AI inferencing hardware.
The Rubin architecture’s cost-effective design has been optimised?for large-scale inference, with computing resources focused on energy efficiency and extensive-context processing. Nvidia says that the platform will allow models to reason across millions of tokens, significantly outperforming current systems.
Due for general release in 2026, Rubin CPX can be utilised for everything from large-scale enterprise applications to next-generation video production and long-context software development.
Visit our website for more news, blog posts, knowledge base articles and information on our wide range of hosting services.
