Headlines

A new Perspective in npj Science of Food lays out a four-pillar design framework for wrapping anti-inflammatory flavonoids in nanoparticles that survive digestion and release their payload directly in the inflamed colon, aiming to turn plant compounds into clinically viable IBD therapies.

A new open-source software package provides fast, modular tools for extracting interpretable cis-regulatory patterns from deep learning models of the genome.

Researchers have developed a physics-informed machine learning framework that predicts the remaining useful life of electric vehicle drive systems with errors under five percent while quantifying uncertainty.

Researchers show that a selective embedding data loading strategy alternating X-ray and CT scans lets deep learning models reach competitive medical image classification accuracy with far less training data.

Researchers in Shanghai have developed a physics-guided neural operator that predicts the levitation air-gap response of EMS maglev trains far faster and more spectrally accurately than existing AI models.

Researchers report that Ailanthus excelsa quassinoids delivered in chitosan nanoparticles suppress STAT3-driven PD-L1 transcription and restore T-cell antitumor activity in triple-negative breast cancer models.

Researchers harnessed a pathology foundation model to read endometrial biopsy slides, predicting IVF live birth outcomes and identifying the receptive phase with AUROCs up to 0.90.

Mount Sinai researchers built a privacy-preserving, on-device forecasting system that uses simple linear models and Fitbit data to predict sedentary bouts in women with chronic pelvic pain disorders, enabling timely just-in-time activity prompts.

A single-step silver-assisted etching process produces vertically aligned silicon nanowires that emit strong visible light through quantum-confined nanocrystals and dramatically enhance Raman scattering signals.

A new calibration framework shows that common benchmarking metrics masked real perturbation-specific signal, and that deep learning models can outperform simple baselines when evaluated properly.

A new multi-stream deep learning framework called MS-AtroNet combines convolutional, state-space, wavelet, and evidential components to classify Alzheimer's MRI slices with record accuracy and reliable uncertainty estimates.

Molecules can be placed in a wide variety of quantum states. How can these states be distinguished in measurements when theoretical models are unreliable? Which excitation process lies behind which measurement signal?

What happens at the nanoscale when information is written to memory? Researchers at KAIST have shown how tiny regions with a new polarization direction form while previously formed regions continue to expand in a promising ferroelectric material. By linking these nanoscale changes to electrical measurements, the team developed a model that captures both processes, offering a basis for designing faster and more stable memory.

A scoping review of 118 studies finds that deep learning models excel at detecting parasites in microscopy images under experimental conditions, but very few have been validated or deployed in real clinical laboratories.

The California State Athletic Commission sent a cease-and-desist letter to a startup that hosted a match between a human and a robot last month, as reported by The New York Times. The fight, which took place on September 18th, pitted a human, Frankie LaPenna, against a humanoid robot owned by a tech startup, Rek, that was being piloted by a human using what the NYT described as a "remote virtual-reality system." Rek says it is the "the humanoid robot fighting league" on its website, and the robot appears to be one from EngineAI but with a Terminator-like head swapped on top. You can watch a replay of the fight on YouTube: The CEO of Rek, … Read the full story at The Verge.

Quantum technologies are considered key technologies of the future. However, research in this field requires not only scientific expertise but sometimes also highly specialized laboratory conditions. Bose–Einstein condensates are a remarkable example, in which atoms move in lockstep, making quantum-mechanical effects visible not only at the level of individual atoms but also on a macroscopic scale. The first Bose–Einstein condensates were observed in ultracold atomic gases at temperatures close to absolute zero. The cooling required to achieve this was technically demanding, costly and restricted to specialized laboratories.

Simulated Patients Before Real Trials A Copenhagen startup has gone on record predicting that drug trials from Biogen

Jessica Hische knew working for Meta might upset some people. But she didn’t anticipate just how angry they would get.

CNBC: Nvidia, Oracle, CoreWeave, and other AI stocks fell after a report that OpenAI's annualized revenue at September's end was $20B less than previously reported  —  Shares of Nvidia, Oracle, CoreWeave and other artificial intelligence names sank lower on Thursday after the market learned more details about OpenAI's revenue.

InvestorPlace - Stock Market News, Stock Advice & Trading Tips Jonathan Rose looks beyond Neuralink at two public companies on his watchlist, explaining what interests him and how their connections and risks differ. The post Before Elon Musk Takes Neuralink Public, Here’s Where to Look for Profits appeared first on InvestorPlace.

© seekingalpha.com. Use of this feed is limited to personal, non-commercial use and is governed by Seeking Alpha's Terms of Use (https://about.seekingalpha.com/terms). Publishing this feed for public or commercial use and/or misrepresentation by a third party is prohibited.

OpenAI has told investors that it hit roughly $50 billion in annualized revenue at the end of September, CNBC confirmed.

OpenAI has told investors that it hit roughly $50 billion in annualized revenue at the end of September, CNBC confirmed.

Researchers in India used Phyllanthus emblica fruit extract as both reducing and stabilizing agent to synthesize crystalline silver nanoparticles whose optical properties can be tuned by precursor concentration.

AI agents arrive on healthcare premisesAI agents are in the news, with warnings that they’re going rogue and may one day bring down the internet or human civilisation. In the meantime it surely won’t be long before AI agents are delivering a big chunk of clinical work, probably well before we understand what their long term impact will be. The latest study edging us closer to an AI doctored future involved a fully “fully on-premise” diagnostic system whereby two AI agents interacted with each other: one took on the role of the physician, the other that of the patient. The agents were good at diagnosing appendicitis, but not so good at diagnosing pneumonia.Nat Med 2026 doi:10.1038/s41591-026-04609-xMulticancer early detection testsPerhaps AI agents will soon relieve humans of providing the NHS health check, and throw in a multicancer early detection test alongside a consultation with no time limit. Findings from the NHS-Galleri...

Artificial intelligence can now solve research problems in mathematics and verify the answers

This summer has featured an amazing run of mathematical discoveries. In May, a prediction first made by Paul Erdős in 1946 about “unit distances” between points in a plane, was disproved. In July, the Jacobian conjecture, an 87-year-old open question in algebraic geometry, was refuted in dimension three. In August, the question of whether a six-dimensional sphere can be given smooth complex coordinates, first asked by Heinz Hopf in 1947, was finally confirmed to be true. Most recently, a finite-time singularity among the solutions to the famous Navier-Stokes equations—a system of equations governing the flow of fluids—was discovered after many decades (in a setting where the fluid is stirred by an external force).

Microsoft’s Surface Laptop Ultra combines Nvidia RTX Spark hardware, up to 128GB of RAM, and professional software support, while lacking high-speed Nvidia connectivity.

The James Webb Space Telescope has detected the farthest "fast radio burst" ever seen, providing more clues about these mysterious explosions.

Researchers have discovered that upconversion nanoparticles can act as optical frequency mixers, generating beat-frequency signals that allow cameras to capture background-free biomedical images even under intense ambient light and deep within tissue.

AI company Anthropic recently announced that its popular system Claude had identified an unusual family of biological molecules. These molecules were enzymes, a type of protein that speeds up chemical reactions in living organisms.

Researchers have shown that concepts erased from text-to-image diffusion models can reappear when structural controls like edge maps and depth maps guide generation, exposing a critical gap in AI safety evaluation.

A study by TU Graz shows that algorithms used by short-form video platforms are particularly effective at keeping people with ADHD engaged.

Plus, news about Roche’s Genentech and Fresenius Kabi. Multiply Labs collects $75M: The robotics biomanufacturing company’s Series B raise will be used to increase its capacity and headcount as it moves ...

Germany gives the green light to Tesla's Full Self-Driving technology but wants to change its name.

Scientists have developed various AI tools to measure people's biological ages. Would you use one?

Magnetic nanoparticles could eventually help doctors both locate and treat ectopic pregnancies, after successfully targeting pregnancy-associated tissue in mice. When a fertilized egg implants inside a fallopian tube instead of the uterus, the growing pregnancy can rupture the tube and trigger life-threatening bleeding. About 98% of these abnormal implantations occur in the fallopian tubes, making [...]

Google’s new Gemini agent gives each coworker agent its own Workspace account, email address, calendar and entry in the company directory, and runs work for days after the laptop closes. Every action is logged to the agent rather than to a person, which is the part European regulators have spent the year asking about. Google […] This story continues at The Next Web

An interpretable AI platform that reads the retinal vasculature predicted preeclampsia and other hypertensive disorders of pregnancy months before clinical onset and transferred successfully to an independent hospital cohort.

Finnish authorities say they cannot conduct a proper environmental assessment after the site has already been changed.

In this edition of STAT's Health Tech: The story of a startup taking its algorithm for breast cancer risk prediction directly to patients, and more.

The first crewed lunar mission in more than a half-century brought back more than 800 gigabytes of data on colorful craters, flashing meteoroids, and much more

RTX Spark laptop prices range from $2,599 to $6,999.99, with higher-cost models offering more unified memory for local AI workloads and demanding professional tasks.

Google has released an experimental note-taking app that can transcribe meetings and audio files entirely offline, as reported earlier by TechCrunch. The app, called Google AI Edge Foresight, is free to use and runs on macOS using the company's on-device EmbeddingGemma 2 model. Similar to AI note-taking apps like Granola and Wispr Flow, Foresight summarizes your meetings while providing a space where you can take your own notes. Google says you can jot down shorthand bullet points during your meeting, and the app will transform them into "polished notes" based on the information in the transcript. Along with transcribing your meetings, For … Read the full story at The Verge.

In early-stage cervical cancer, radical hysterectomy is an established treatment. The operation can be performed either through open surgery or robot-assisted minimally invasive surgery. Until now, there have been limited data from large randomized studies on differences in complications and recovery between the methods.

Vendors are pitching AI as a cure-all for legacy IT systems, but is it?

Bloomberg: Sources: the IPO of Nvidia-backed data center provider Firmus is uncertain after failing to attract adequate support, underscoring concerns over AI funding  —  The sudden plunge in demand for an Nvidia-backed data center company's initial public offering is revealing fresh cracks in the AI funding boom.

There’s a big disconnect between the rate at which IT is deploying various kinds of AI-generated output and the speed with which it can roll those changes back when things go wrong, warns a new report from Fleet Device Management. It’s almost as if the rush to embrace AI has eclipsed the need to manage its deployment effectively. The research, based on a survey of more than 250 enterprise IT practitioners managing Apple devices, is available in full via the company’s website and echoes similar concerns I’ve heard from others in the space. I spoke with Fleet founder and CEO Mike McNeil to get his take on the disconnect.  What’s happening in the enterprise First, some of the stats gathered in the survey: 86% of respondents allow AI-written output to reach production devices following some review. 61% used AI to

A new multimodal deep learning framework called GeoEPred combines protein language model semantics with predicted 3D structures to more accurately identify the virulence effector proteins secreted by Gram-negative bacteria.

VEIR has raised $110 million in a Series C round. The Massachusetts company makes superconducting power cables for AI data centres. Matter Venture Partners and Tyche Partners co-led the round, which VEIR called oversubscribed. LG Technology Ventures, Gates Frontier, Sabanci Climate Ventures and Hui Capital joined as new investors. Existing backers also took part, including […] This story continues at The Next Web

Futurism says 19 Facebook accounts containing graphic AI-generated videos depicting violence against children remained accessible more than 24 hours after being reported to Meta.

Google is launching a "universal" Gemini AI agent that can work across apps and devices in the background. The tool, announced as part of the Gemini at Work event on Thursday, will be available within the Gemini Enterprise app, allowing users to chat with the Gemini agent and assign it tasks from a single interface. In addition to having Gemini work directly inside Workspace apps like Gmail, Drive, Docs, Sheets, Calendar, and others, users can interact with the agent from their mobile device, desktop computer, the web, and third-party apps like Slack or Microsoft 365. Since it runs in the cloud, Gemini will maintain the same context across … Read the full story at The Verge.

Colin Demarest / Axios: Gallatin AI, which uses AI to digitize and streamline military logistics data, raised a $50M Series A, bringing its total funding to $70M  —  Gallatin AI raised a $50 million Series A, with defense-tech frequent fliers 8VC and Silent Ventures pitching in, CEO Woody Glier told Axios.

By combining millisecond-resolution X-ray scatter microscopy with polarized light imaging, researchers have shown that plate-like and rod-like nanoparticles respond in fundamentally different ways to the vortex instabilities of transitional Taylor–Couette flow.

No more AI training in Russia? Destiny of two Nvidia-powered Russian supercomputers is unknown as drones hit Yandex data center in Sasovo.

Google has launched Playground, an experimental platform that uses natural-language prompts to help users create, customise and share browser games without coding experience.

Ivan Mehta / TechCrunch: Google debuts Google AI Edge Foresight, a local macOS note-taking app for meetings that uses its 740M-parameter EmbeddingGemma 2 model and Gemma 4 assistant  —  Earlier in April, Google released an AI-powered dictation tool that worked with local models.  Now, the same team has released …

Carl Franzen / VentureBeat: Google Cloud unveils the Gemini agent to handle multi-day enterprise workflows across Workspace, Microsoft 365, and Slack, with support for multiple AI models  —  Google is unveiling a new universal AI agent for the workplace that can research information, create documents, write code …

Researchers have built an interpretable machine-learning framework that predicts sovereign default as a multi-class event, identifying which creditors a government is likely to fail to pay.

Upscale AI, a Santa Clara networking startup backed by Nvidia, launched a platform on Thursday that lets data centres link AI chips made by different suppliers, Reuters reported. The product, called Token Fabric, was revealed in a company announcement today. Upscale said it is built on open standards and works with graphics chips, custom AI chips, […] This story continues at The Next Web

OpenAI CEO Sam Altman has warned against treating AI as a religious authority after xAI's Grok chatbot described Jesus as the Son of God in response to a user request.

The system can be deployed in both field and urban environments, Rosoboronexport said

Researchers at Shanghai Jiao Tong University have developed a fault diagnosis framework that uses recurrent attentional reinforcement learning to adaptively search for the most informative temporal fragments in machinery sensor data, achieving 76.4 percent accuracy on rolling bearings and 100 percent on a hydraulic system.

With AI handling admin work, graduates could end up managing AI agents, but what happens to middle managers?

Robots are becoming more specialized, lightweight, and adaptable. From industrial robots and collaborative robots to humanoid and autonomous

At Gemini at Work 2026, Google Cloud today announced the “Gemini agent” for enterprise customers. more…

Researchers developed a chitosan-coated, manganese-doped bismuth ferrite nanoplatform that responds to acidic tumor conditions, supports T2 MRI monitoring, and combines chemodynamic therapy with ultrasound-triggered piezodynamic therapy. In subcutaneous U87 glioma-derived tumors in mice, the treatment markedly reduced tumor volume and prolonged survival; in vitro tests also showed pH-responsive MRI behavior, ROS generation, and enhanced transport across a BBB model.

Superconductors, materials that can carry electric current with an electrical resistance of zero, have proved to be promising for the development of various technologies, including medical devices, particle accelerators and quantum computers. Studying these materials could help researchers uncover new physical states that could be useful for specific applications.

Full-stack safety solution for physical AI is being used by robotics companies.

A secondary analysis of the RACC trial assesses perioperative safety and early quality of life after robot-assisted vs open radical hysterectomy in patients with early-stage cervical cancer.

A new commentary argues that lung nodule localisation is entering a precision era, as stabilised fluorescent matrices aim to solve the diffusion problem that undermines surgical margins in robotic sublobar resection.

“What I finally decided was the organizing principle was that psychiatry aims at reduction of symptoms, but not restoration of the full human experience that makes life most worthwhile.”‒ Linus Abrams The post #404 ‒ Mental health beyond neurotransmitters: the role of hormones in psychiatry, why symptom reduction isn’t enough, and the future of psychedelic therapies | Linus Abrams, M.D. appeared first on Peter Attia MD.

SpaceX reportedly intends to borrow $40 billion to procure 360,000 Nvidia Rubin AI accelerators, infrastructure.

Researchers have filmed ultrafast spin-driven charge currents with simultaneous femtosecond and nanometer resolution using fiber-coupled terahertz emission nanoscopy, explaining why near-field tips can detect in-plane currents through their out-of-plane field signatures.

Philosophers propose that autonomous vehicles should be trained as digital duplicates of wise local drivers, endowed with a functional version of Aristotle's practical wisdom to handle everyday traffic coordination.

Amazon used the Atlas union monitoring systems at sites around the world, including in the UK.

A study of 201 cooking robot users identifies four factors associated with trust and examines how perceived risk influences adoption intentions.

Seoul’s push to expand the use of autonomous buses and taxis is raising questions about the future of transport workers, with unionized drivers calling for stronger safeguards and a greater say in a transition that could eventually threaten their jobs. While city officials say the technology will initially help address driver shortages rather than displace workers, union representatives want clear answers about their future job security, passenger safety and responsibility in the event of an accident. “Self-driving technology should be limited to supporting human drivers, instead of replacing them,” a Seoul bus union official surnamed Yoo told The Korea Times recently. “Operating without humans is neither feasible nor desirable.” He said drivers’ responsibilities extend beyond simply steering and braking to responding to passenger inquiries, preventing door-related injuries and handling unexpected incidents. He supported technology that can assist them with risks such as drowsy driving

The story is a collaboration between MIT Technology Review and Aventine, a non-profit research foundation that creates and supports content about how technology and science are changing the way we live. A robot shaped like a human—white with a black head and torso—has been popping up on video feeds. Perhaps you’ve seen it dance or…

Why scaling physical AI requires more than investment, ambition and impressive pilot projects.

A tech start-up’s stunt in San Francisco tested the boundaries of combat sport safety and regulations.

A new deep learning framework called MedSegNet-AXU segments multiple sclerosis lesions from 3D MRI with a 98.58 percent Dice score and uses radiomics and graph neural networks to predict sensory, motor, and visual impairment.

Eurasia Group’s Dan Wang noted that the fundraising is evidence of investors' continued interest in backing Manus, despite the fall out from the Meta incident. Read more: AI start-up Manus confirms fundraising of $500m post Meta’s failed acquisition

Researchers in Vietnam have built a hybrid vision-language AI that diagnoses grapevine diseases with 98.7 percent accuracy while explaining its reasoning and resisting the background overfitting that cripples conventional CNNs in real fields.

A startup didn't want to wait for doctors and insurers to get behind its breast-cancer risk prediction AI. So it's going directly to patients.

Seoul’s push to expand the use of autonomous buses and taxis is raising questions about the future of transport workers, with unionized drivers calling for stronger safeguards and a greater say in a transition that could eventually threaten their jobs. While city officials say the technology will initially help address driver shortages rather than displace workers, union representatives want clear answers about their future job security, passenger safety and responsibility in the event of an accident. “Self-driving technology should be limited to supporting human drivers, instead of replacing them,” a Seoul bus union official surnamed Yoo told The Korea Times recently. “Operating without humans is neither feasible nor desirable.” He said drivers’ responsibilities extend beyond simply steering and braking to responding to passenger inquiries, preventing door-related injuries and handling unexpected incidents. He supported technology that can assist them with risks such as drowsy driving

AI agents don’t go rogue. That’s something only humans do. Nevertheless, a New York Times article – representative

Google has, to date, watermarked more than 180bn images and videos, it said. Read more: Google makes SynthID AI detector publicly available

Researchers at Heidelberg University Hospital developed a machine learning model combining ultrasound, mammography, MRI and clinical data to predict pathological complete response after neoadjuvant chemotherapy in breast cancer, achieving an AUROC of 0.76.

Starting today, Google is making its SynthID Detector tool available to everyone across the globe in English. This tool can be used to find out if an image, video, or audio file was made with AI from Google, OpenAI, Nvidia, or Kakao. Soon, the SynthID Detector will also detect media that was created with Apple's AI tools. For its part, Google says it's watermarked over 180 billion images and videos and "240,000 years of audio content" with SynthID since it was launched in 2023. SynthID embeds imperceptible watermarks into images, video, and audio, and the SynthID Detector basically...

Using nanoscale vibrational spectroscopy inside carbon nanotubes, researchers have directly observed water losing its hydrogen-bond network under extreme confinement.

The deal is handing the AI agent developer fresh capital after Beijing forced Meta to unwind its roughly $2 billion acquisition.