Only world models respond to the user’s input as they navigate around the world by moving the camera, or interacting with people and objects it contains, rather than just interpreting prompts to decide what video should be generated.
Using this method, the entire world is continuously generated, frame-by-frame, based on the model’s internal understanding of how the environment and objects should behave.
This method allows the creation of highly flexible, realistic and unique environments. Imagine a video game world, for example, where literally anything can happen. The possibilities aren’t limited to situations and choices that have been written into the code by a game programmer, because the model generates visuals and sounds to match any choice the player makes.
A new advance from Carnegie Mellon University researchers could reshape how clinicians identify the brain regions responsible for drug-resistant epilepsy. Surgery can be a life-changing option for millions of epilepsy patients worldwide, but only if physicians can accurately locate the epileptogenic zone, the area where seizures originate.
Bin He, professor of biomedical engineering, and his team have developed a unified, machine learning-based approach called spatial-temporal-spectral imaging (STSI) to assist. It is the first technology capable of analyzing every major type of epileptic brain signal within a single computational framework.
Their work, published in PNAS, presents a technical breakthrough and promising new direction for noninvasive presurgical planning.
Picus Security explains why relying on LLM-generated attack scripts is risky and how an agentic approach maps real threat intel to safe, validated TTPs. Their breakdown shows how teams can turn headline threats into reliable defense checks without unsafe automation.
Two malicious extensions on Microsoft’s Visual Studio Code Marketplace infect developers’ machines with information-stealing malware that can take screenshots, steal credentials, crypto wallets, and hijack browser sessions.
The marketplace hosts extensions for the popular VSCode integrated development environment (IDE) to extend functionality or add customization options.
The two malicious extensions, called Bitcoin Black and Codo AI, masquerade as a color theme and an AI assistant, respectively, and were published under the developer name ‘BigBlack.’
Google is introducing in the Chrome browser a new defense layer called ‘User Alignment Critic’ to protect upcoming agentic AI browsing features powered by Gemini.
Agentic browsing is an emerging mode in which an AI agent is configured to autonomously perform for the user multi-step tasks on the web, including navigating sites, reading their content, clicking buttons, filling forms, and carrying out a sequence of actions.
User Alignment Critic is a separate LLM model isolated from untrusted content that acts as a “high-trust system component.”
In a significant advance for astronomy, researchers from the Breakthrough Listen initiative, working in collaboration with NVIDIA and utilizing their system on the SETI Institute’s Allen Telescope Array (ATA) in California, have improved the process for detecting Fast Radio Bursts (FRBs). Their newly developed artificial intelligence system outperforms existing methods, operating hundreds of times faster than current pipelines while maintaining accuracy.
Detailed in the peer-reviewed journal Astronomy & Astrophysics, the new system operates on NVIDIA’s Holoscan platform, designed to process massive streaming datasets in real-time. Traditionally, FRB detection requires “dedispersion” — searching through thousands of possible signal parameters to correct for frequency-dependent time delays. The new end-to-end AI architecture eliminates that bottleneck, analyzing signals in real time and transforming how astronomers search for transient and potentially artificial signals from space.
The performance gains are notable. At the ATA, the state-of-the-art pipeline currently takes approximately 59 seconds to process 16.3 seconds of observational data — nearly four times slower than real-time. The new AI-driven system performs the same task 600 times faster, operating over 160 times faster than real-time.
A radically miniaturized brain implant called BISC is redefining what’s possible in human–computer interaction, offering a paper-thin, wireless, high-bandwidth link directly to the brain.
With over 65,000 electrodes and unprecedented data throughput, it enables advanced AI decoding of thoughts, intentions, and sensory experiences while remaining minimally invasive.