Top Researchers

UK Creates First Cabinet-Level AI Minister Under PM Andy Burnham

Fei-Fei Li's World Labs has acquired SceniX, a robotics startup focused on closing the real-to-sim gap, bringing aboard Yunzhu Li's team to train and evaluate robots in high-fidelity simulation. The…

208 themes mapped1231 sourcesActivity 99/100

Next signals to watch

Daily brief

Five signals shaping AI today

Live map

A 60-second overview before you explore the evidence graph.

Daily brief · Jul 27

Five signals shaping AI today

UK Creates First Cabinet-Level AI Minister Under PM Andy Burnham

4 sources · 1 corroborated

New UK Prime Minister Andy Burnham appointed Kanishka Narayan as Minister of State for AI with full Cabinet attendance, creating the UK's first-ever cabinet-level AI ministerial position. [2][3]

Google DeepMind CEO Demis Hassabis publicly congratulated Narayan on the appointment, calling it "great news for the UK AI ecosystem." [1][4]

World Labs acquires SceniX to expand spatial intelligence into robotics

5 sources · 1 corroborated

World Labs, Fei-Fei Li's spatial intelligence company, has acquired SceniX, a robotics-focused startup built to close the real-to-sim gap for robot learning, with SceniX's team joining World Labs to train and evaluate robots in high-fidelity simulation. [2][3]

Fei-Fei Li framed the acquisition as an expansion beyond generating and perceiving worlds, stating that spatial intelligence is about interacting with the world — positioning robotics as the next frontier for World Labs' work on spatial intelligence. [1][4][5]

The SceniX team joining World Labs is led by Yunzhu Li (@YunzhuLiYZ), along with collaborators identified by Fei-Fei Li as @fast_sploosh and @xhsonny. [2][3]

NVIDIA expands world models, simulation tools, and neural rendering at SIGGRAPH…

5 sources · 1 corroborated

At SIGGRAPH 2026, NVIDIA expanded its Agent Toolkit with new Omniverse libraries that enable AI agents to automatically inspect, validate, and prepare simulation-ready 3D worlds, giving agents the ability to build and configure environments for physical AI workflows. [2][3][4]

NVIDIA's SIGGRAPH keynote featured research and engineering leaders Neil Ashton, Edward Liu, and Ming-Yu Liu presenting advances in neural rendering, world models, and simulation methods, with NVIDIA CEO Jensen Huang stating the goal is "to create virtual worlds that behave with the fidelity and realism of the physical world" across domains including robotics, industrial design, and autonomous systems. [1]

NVIDIA revealed technical details of DLSS 5 at SIGGRAPH 2026, introducing a third AI performance category called real-time neural rendering, based on a compact diffusion transformer model distilled from large core neural networks; the system takes a standard color buffer, motion vectors, and internal engine data to enrich the final image, improving global illumination, subsurface scattering, hair lighting, and fabric physics while preserving the scene's geometric structure and artistic intent. [5]

DLSS 5 is integrated via the standard NVIDIA Streamline package, with an official release scheduled for fall 2026; confirmed supporters include Bethesda, Capcom, and Ubisoft, with first titles being Resident Evil Requiem, Starfield, and Assassin's Creed. [5]

NVIDIA Launches Cosmos 3 Edge 4B-Parameter On-Device World Model

6 sources · 1 corroborated

NVIDIA announced Cosmos 3 Edge at SIGGRAPH on July 20, an open frontier world model with approximately 4 billion parameters built on a Nemotron-based reasoner architecture, designed to run entirely on-device (edge GPUs such as NVIDIA Jetson Thor) without requiring cloud compute round-trips. The model covers perception, prediction, and action for physical AI applications including robotics and autonomous driving. [1][2][3][4][5][6]

Cosmos 3 Edge is openly available and received Day-0 support from vLLM and vLLM Omni at launch, making it directly deployable on local edge hardware. NVIDIA positioned it as bringing "frontier world model capabilities" to on-device deployment, a category previously requiring cloud-scale infrastructure. [3][5]

The SIGGRAPH keynote featuring Cosmos 3 Edge was presented by NVIDIA AI research and engineering leaders Neil Ashton, Edward Liu, and Ming-Yu Liu, covering neural rendering techniques, world models, and simulation methods for AI. NVIDIA CEO Jensen Huang stated the goal is "to create virtual worlds that behave with the fidelity and realism of the physical world," applicable to games, cinema, robotics, and factory digital twins. [3]

AlphaFold used to redesign gene-editing proteins and reduce off-target effects

2 sources · 1 corroborated

A study published in Nature describes researchers modifying AlphaFold — originally designed for protein structure prediction — to model enzyme-DNA complexes and identify the structural features of gene-editing (CRISPR) Cas proteins that are responsible for off-target edits, where the system edits an unintended DNA sequence rather than the target gene. [1][2]

Using AlphaFold's structural insights, researchers identified key regions within the Cas protein responsible for off-target activity and then modified those regions to reduce such errors; this approach addresses a fundamental safety challenge in gene-editing therapies, where editing large numbers of cells makes even low-probability off-target events statistically inevitable. [2]

The work represents a downstream application of AlphaFold beyond its original structure-prediction use case, using it as a design tool to engineer safer gene-editing proteins — complementing existing safety approaches such as selecting guide RNA sequences that have minimal similarity to other genomic locations. [1][2]