Research News
NYU Tandon researchers develop new AI system that leverages standard security cameras to detect fires in seconds, and could transform emergency response
Fire kills nearly 3,700 Americans annually and destroys $23 billion in property, with many deaths occurring because traditional smoke detectors fail to alert occupants in time.
Now, the NYU Fire Research Group at NYU Tandon School of Engineering has developed an artificial intelligence system that could significantly improve fire safety by detecting fires and smoke in real-time using ordinary security cameras already installed in many buildings.
Published in the IEEE Internet of Things Journal, the research demonstrates a system that can analyze video footage and identify fires within 0.016 seconds per frame—faster than the blink of an eye—potentially providing crucial extra minutes for evacuation and emergency response. Unlike conventional smoke detectors that require significant smoke buildup and proximity to activate, this AI system can spot fires in their earliest stages from video alone.
"The key advantage is speed and coverage," explained lead researcher Prabodh Panindre, Research Associate Professor in NYU Tandon’s Department of Mechanical and Aerospace Engineering (MAE). "A single camera can monitor a much larger area than traditional detectors, and we can spot fires in the initial stages before they generate enough smoke to trigger conventional systems."
The need for improved fire detection technology is evident from concerning statistics: 11% of residential fire fatalities occur in homes where smoke detectors failed to alert occupants, either due to malfunction or the complete absence of detectors. Moreover, modern building materials and open floor plans have made fires spread faster than ever before, with structural collapse times significantly reduced compared to legacy construction.
The NYU Tandon research team developed an ensemble approach that combines multiple state-of-the-art AI algorithms. Rather than relying on a single AI model that might mistake a red car or sunset for fire, the system requires agreement between multiple algorithms before confirming a fire detection, substantially reducing false alarms, a critical consideration in emergency situations.
The researchers trained their models by building a comprehensive custom image dataset representing all five classes of fires recognized by the National Fire Protection Association, from ordinary combustible materials to electrical fires and cooking-related incidents. The system achieved notable accuracy rates, with the best-performing model combination reaching 80.6% detection accuracy.
The system incorporates temporal analysis to differentiate between actual fires and static fire-like objects that could trigger false alarms. By monitoring how the size and shape of detected fire regions change over consecutive video frames, the algorithm can distinguish between a real, growing fire and a static image of flames hanging on a wall. "Real fires are dynamic, growing and changing shape," explained Sunil Kumar, Professor of MAE. "Our system tracks these changes over time, achieving 92.6% accuracy in eliminating false detections."
The technology operates within a cloud-based Internet of Things architecture where multiple standard security cameras stream raw video to servers that perform AI analysis. When fire is detected, the system automatically generates video clips and sends real-time alerts via email and text message. This design means the technology can be implemented using existing CCTV infrastructure without requiring expensive hardware upgrades, an important advantage for widespread adoption.
This technology can be integrated into drones or unmanned aerial vehicles to search for wildfires in remote forested areas. Early-stage wildfire detection would buy critical hours in the race to contain and extinguish them, enabling faster dispatch of resources, and prioritized evacuation orders that dramatically reduce ecological and property loss.
To improve the safety of firefighters and assist during fire response, the same detection system can be embedded into the tools firefighters already carry: helmet cameras, thermal imagers, and vehicle-mounted cameras, as well as into autonomous firefighting robots. In urban areas, UAVs integrated with this technology can help the fire service in performing 360-degree size-up, especially when fire is on higher floors of high-rise structures.
“It can remotely assist us in confirming the location of the fire and possibility of trapped occupants,” said Capt. John Ceriello from the Fire Department of New York City.
Beyond fire detection, the researchers note their approach could be adapted for other emergency scenarios such as security threats or medical emergencies, potentially expanding how we monitor and respond to various safety risks in our society.
In addition to Panindre and Kumar, the research team includes Nanda Kalidindi (’18 MS Computer Science, NYU Tandon), Shantanu Acharya (’23 MS Computer Science, NYU), and Praneeth Thummalapalli (’25 MS Computer Science, NYU Tandon).
P. Panindre, S. Acharya, N. Kalidindi and S. Kumar, "Artificial Intelligence-Integrated Autonomous IoT Alert System for Real-Time Remote Fire and Smoke Detection in Live Video Streams," in IEEE Internet of Things Journal, doi: 10.1109/JIOT.2025.3598979.
Coral-inspired pill offers a new window into the hidden world of the gut
In the depths of the ocean, marine corals have evolved intricate, porous structures that shelter diverse microbial communities.
Now, researchers have borrowed this biological blueprint to create an ingestible pill that can sample bacteria from one of the most inaccessible regions of the human body: the small intestine.
The CORAL (Cellularly Organized Repeating Lattice) capsule, developed by Khalil Ramadi – assistant professor of bioengineering at NYU Tandon School of Engineering and NYU Abu Dhabi (NYUAD) – and NYUAD collaborators, promises the first passive, non-invasive way to collect microbes from the upper digestive tract. Once swallowed, the device physically traps bacteria as it travels naturally through the digestive system before exiting the body.
In a study published in Device, the team demonstrated that their coral-inspired device provides a more comprehensive picture of the small intestine's bacterial landscape than traditional stool samples, the current gold standard for microbiome research.
"Fecal samples, though easy to collect, do not accurately represent the microbial communities in distinct regions of the gut," said Ramadi, who directs the Laboratory for Advanced Neuroengineering and Translational Medicine at NYUAD.
While the gut microbiome has been linked to everything from immune disorders to mental health, most stool-sample studies primarily reflect bacteria from the large intestine and miss the unique microbial communities of the small intestine. This matters because the small intestine is where much of the critical action occurs. As the body's largest mucosal surface, it hosts a high density of receptors, immune cells, and neurons, making it a crucial site for host-microbiome interactions.
Recent research suggests that various diseases — including immune disorders, metabolic diseases, and endocrine diseases — may actually originate in the gut, with distinct microbial populations in the small intestine playing essential roles in metabolism and immunity that differ significantly from those in the colon.
"The CORAL capsule captures bacteria that are otherwise not accessible, addressing a significant blind spot in microbiome science," said Aashish Jha, Assistant Professor of Biology at NYUAD and the paper’s co-senior author. "Understanding these upstream microbial communities could be key to early disease detection and developing more targeted therapeutic interventions."
The capsule's design mimics marine corals using mathematically defined structures called Triply Periodic Minimal Surfaces (TPMS). These create a maze-like network of channels with pore sizes optimized to trap bacteria while allowing safe passage through the digestive tract.
Unlike existing microbiome sampling devices that rely on magnets, mechanical actuators, or electronic components, CORAL operates entirely passively. The capsule is fabricated in a single 3D printing step and contains no moving parts, making it potentially scalable for widespread use. A special coating ensures the device only begins sampling once it reaches the small intestine, avoiding contamination from stomach acid.
"We designed CORAL to be as simple as possible, no batteries or electronics, just a mathematically precise structure that uses the gut's natural movement to sample bacteria," said Hanan Mohammed, lead author of the study and Research Associate at NYUAD. "It gives us access to bacterial communities that have been invisible to researchers until now."
In animal studies, CORAL successfully captured distinct bacterial populations from the small intestine that differed significantly from fecal samples. The capsule samples collected higher levels of beneficial bacteria like Lactobacillus, which thrives in the upper gut's lower pH environment, while intentionally missing bacteria typically found in the large intestine.
This work represents part of Ramadi's broader mission to change how we can diagnose and treat diseases through the gut. His work involves developing "electroceuticals" — ingestible electricals rather than pharmaceutical interventions — that can diagnose and treat conditions from immune disorders to metabolic diseases by leveraging the body's natural neural pathways.
The team envisions translating CORAL to eventual human use by scaling the capsule from its current tiny dimensions to standard pill size. Before human trials could begin, researchers would need to develop reliable retrieval methods (potentially using magnetic detection or other identification techniques) and conduct extensive safety testing to ensure the device poses no risk to patients. The team is continuing this work in the lab and actively working to commercialize this technology through the HealthX program at StartAD and the Abu Dhabi Department of Health.
In addition to Ramadi, Jha, and Mohammed, the paper's co-authors are Sadaf Usmani, Brij Bhushan, Anique Ahmad, Oraib Al-Ketan, Ahmed A. Shibl, Maylis Boitet, and Heba Naser, all at NYU Abu Dhabi, and Devjoy Dev at both NYU Abu Dhabi and NYU Tandon.
Comments on CORAL:
This study makes a major contribution to microbiome research. By allowing precise and minimally invasive access to the small intestine, the CORAL capsules enable characterization of microbial communities that have until now remained largely out of reach. This breakthrough provides an essential tool for advancing basic science and for shaping the next generation of microbiome-based diagnostics and therapies.”
María Rodríguez Aburto, Ph.D.
Senior Lecturer, ERC-funded investigator
Dept. of Anatomy & Neuroscience, APC Microbiome Ireland
University College Cork
The relationship between the small intestinal microbiome and immune and digestive function is woefully understudied due to difficulties in sampling the environment in situ. The CORAL device is a key technological advance that enables sampling of the small intestinal microbiome to probe its importance in health and disease.”
Mark Mimee, Ph.D.
Assistant Professor of Microbiology
Committee on Molecular Metabolism and Nutrition
The University of Chicago
Passive intestinal microbiome sampling using an ingestible device with tortuous lattices
Mohammed, Hanan et al. Device, Volume 0, Issue 0, 100904
Large language models can execute complete ransomware attacks autonomously, NYU Tandon research shows
Criminals can use artificial intelligence, specifically large language models, to autonomously carry out ransomware attacks that steal personal files and demand payment, handling every step from breaking into computer systems to writing threatening messages to victims, according to new research from NYU Tandon School of Engineering.
The study serves as an early warning to help defenders prepare countermeasures before bad actors adopt these AI-powered techniques.
A simulation malicious AI system developed by the Tandon team carried out all four phases of ransomware attacks — mapping systems, identifying valuable files, stealing or encrypting data, and generating ransom notes — across personal computers, enterprise servers, and industrial control systems.
This system, which the researchers call “Ransomware 3.0," became widely known recently as "PromptLock," a name chosen by cybersecurity firm ESET when experts there discovered it on VirusTotal, an online platform where security researchers test whether files can be detected as malicious.
The Tandon researchers had uploaded their prototype to VirusTotal during testing procedures, and the files there appeared as functional ransomware code with no indication of their academic origin. ESET initially believed they found the first AI-powered ransomware being developed by malicious actors. While it is the first to be AI-powered, the ransomware prototype is a proof-of-concept that is non-functional outside of the contained lab environment.
"The cybersecurity community's immediate concern when our prototype was discovered shows how seriously we must take AI-enabled threats," said Md Raz, a doctoral candidate in the Electrical and Computer Engineering Department who is the lead author on the Ransomware 3.0 paper the team published publicly. "While the initial alarm was based on an erroneous belief that our prototype was in-the-wild ransomware and not laboratory proof-of-concept research, it demonstrates that these systems are sophisticated enough to deceive security experts into thinking they're real malware from attack groups."
The research methodology involved embedding written instructions within computer programs rather than traditional pre-written attack code. When activated, the malware contacts AI language models to generate Lua scripts customized for each victim's specific computer setup, using open-source models that lack the safety restrictions of commercial AI services.
Each execution produces unique attack code despite identical starting prompts, creating a major challenge for cybersecurity defenses. Traditional security software relies on detecting known malware signatures or behavioral patterns, but AI-generated attacks produce variable code and execution behaviors that could evade these detection systems entirely.
Testing across three representative environments showed both AI models were highly effective at system mapping and correctly flagged 63-96% of sensitive files depending on environment type. The AI-generated scripts proved cross-platform compatible, operating on (desktop/server) Windows, Linux, and (embedded) Raspberry Pi systems without modification.
The economic implications reveal how AI could reshape ransomware operations. Traditional campaigns require skilled development teams, custom malware creation, and substantial infrastructure investments. The prototype consumed approximately 23,000 AI tokens per complete attack execution, equivalent to roughly $0.70 using commercial API services running flagship models. Open-source AI models eliminate these costs entirely.
This cost reduction could enable less sophisticated actors to conduct advanced campaigns previously requiring specialized technical skills. The system's ability to generate personalized extortion messages referencing discovered files could increase psychological pressure on victims compared to generic ransom demands.
The researchers conducted their work under institutional ethical guidelines within controlled laboratory environments. The published paper provides critical technical details that can help the broader cybersecurity community understand this emerging threat model and develop stronger defenses.
The researchers recommend monitoring sensitive file access patterns, controlling outbound AI service connections, and developing detection capabilities specifically designed for AI-generated attack behaviors.
The paper's senior authors are Ramesh Karri — ECE Professor and department chair, and faculty member of Center for Advanced Technology in Telecommunications (CATT) and NYU Center for Cybersecurity — and Farshad Khorrami — ECE Professor and CATT faculty member. In addition to lead author Raz, the other authors include ECE Ph.D. candidate Meet Udeshi; ECE Postdoctoral Scholar Venkata Sai Charan Putrevu and ECE Senior Research Scientist Prashanth Krishnamurthy.
The work was supported by grants from the Department of Energy, National Science Foundation, and from the State of New York via Empire State Development's Division of Science, Technology and Innovation.
Raz, Md, et al. “Ransomware 3.0: Self-Composing and LLM-Orchestrated.” arXiv.Org, 28 Aug. 2025, doi.org/10.48550/arXiv.2508.20444.
NYU Tandon-led team develops new fabrication technique that opens door to new materials for quantum technologies
Researchers have demonstrated a new fabrication approach that enables the exploration of a broader range of superconducting materials for quantum hardware.
The study, published in Applied Physics Letters and selected as a “Featured” article, addresses a long-standing challenge: many promising superconductors, such as transition metal nitrides, carbides, and silicides, are difficult to pattern into functional devices using conventional chemistry-based methods.
By showing that physical patterning provides a viable alternative, the study paves the way to evaluate and harness these materials for high-performing quantum technologies.
The team, led by NYU Tandon professor Davood Shahrjerdi, demonstrated that one such technique, called low-energy ion beam etching (IBE), can be used to fabricate high-performing quantum devices. They validated the approach using niobium, a well-studied superconductor, and benchmarked the resulting devices against state-of-the-art counterparts made with conventional chemistry-based methods, showing comparable performance.
Quantum computers have the potential to tackle problems that are intractable for today's machines, with applications in drug discovery, cryptography, and financial modeling.
"Realizing this promise requires components that can preserve fragile quantum states long enough to perform complex calculations," said Shahrjerdi. "That means building ever more perfect hardware to reduce errors and improve the fault tolerance of quantum systems."
The team's demonstration advances this broader goal by expanding the superconducting material toolkit for device fabrication.
"Fabricating devices with materials-agnostic techniques expands the design space for quantum hardware to under-explored materials, which could catalyze advancements in the scaling of quantum information systems to greater size and functionality," said Dr. Matthew LaHaye, a research physicist at the Air Force Research Laboratory (AFRL) and a collaborator on the project.
To put this approach to the test, Ph.D. students Miguel Manzo-Perez and Moeid Jamalzadeh, co-lead authors of the study, designed superconducting resonators and developed fabrication protocols that combined electron-beam lithography with IBE. They deposited thin niobium films on silicon substrates and patterned them into superconducting resonators, completing the entire process at the NYU Nanofabrication Cleanroom (NYU Nanofab), the first academic cleanroom in Brooklyn.
"NYU Nanofab is equipped with state-of-the-art tools and a strategic focus on enabling the fabrication of advanced devices from quantum materials and superconductors," said Shahrjerdi, the inaugural Faculty Director. "In addition to advancing academic research, it also serves as the prototyping facility of the Northeast Regional Defense Technology (NORDTECH) Hub, with the mission to support lab-to-fab transitions in superconducting quantum technologies."
Next, the devices were shipped to AFRL, where Booz Allen Hamilton contractors Christopher Nadeau and Man Nguyen tested them at temperatures near absolute zero. The quantum resonators demonstrated high performance, confirming the feasibility of the IBE-based fabrication approach for realizing low-loss quantum hardware.
Loss is a critical measure of hardware quality, with lower values indicating more perfect superconducting devices.
In addition to Shahrjerdi, LaHaye, Manzo-Perez, Jamalzadeh, Nadeau, and Nguyen, the other co-authors of the paper include Alexander Madden of Booz Allen Hamilton; Iliya Shiravand of NYU Tandon; Kim Kisslinger and Xiao Tong of Brookhaven National Laboratory; Kasra Sardashti of the University of Maryland; and Michael Senatore of the Air Force Research Laboratory.
NYU Tandon and AFRL Rome collaborate under a Cooperative Research and Development Agreement (24-RI-CRADA-09) and are supported by funding from the Microelectronics Commons through the Northeast Defense Technology Hub project entitled "Improved Materials for Superconducting Qubits with Scalable Fabrication."
Miguel Manzo-Perez, Moeid Jamalzadeh, Man Nguyen, Christopher Nadeau, Alexander Madden, Iliya Shiravand, Kim Kisslinger, Xiao Tong, Kasra Sardashti, Michael Senatore, Matthew LaHaye, Davood Shahrjerdi; Physical patterning of high-Q superconducting niobium resonators via ion beam etching. Appl. Phys. Lett. 1 September 2025; 127 (9): 092601. https://doi.org/10.1063/5.0278956
Light pills could transform understanding of how the gut controls the body
Scientists have long struggled with how to study the gut's vast nervous system — often called the body's 'second brain' — without damaging it. Current research methods are invasive and often require complex surgeries that make it difficult to study normal gut function.
"If you look at how we do any study trying to map neural function in the gut, it is all extremely crude," said Khalil Ramadi, a NYU researcher who has developed a new approach to this challenge. "We just don't have good tools for it."
A team led by Ramadi — assistant professor of bioengineering at NYU Tandon School of Engineering and Director of the Laboratory for Advanced Neuroengineering and Translational Medicine at NYU Abu Dhabi (NYUAD) — has created ingestible devices called ICOPS (Ingestible Controlled Optogenetic Stimulation) that deliver targeted light stimulation directly to the gut.
The technology allows researchers to precisely illuminate specific regions of the intestinal tract, activating specific nerve cells. It could be used to observe how those cells control digestion, for example, and reveal new targets for treating conditions like gastroparesis, where the stomach empties too slowly, or metabolic diseases and eating disorders. The approach represents a dramatic improvement over current methods, which typically involve invasive surgical procedures to implant optical fibers .
The device enables optogenetics, a technique that makes specific cells light-sensitive. Scientists first modify target neurons to respond to light stimulation, then the patient swallows the LED-equipped pill.
"You can go in, transfect a certain subset of cells to be light sensitive, and then swallow this light pill whenever you want to activate those cells," Ramadi explained.
In a paper published in Advanced Materials Technologies, the researchers demonstrate how these devices could control the enteric nervous system — the network of neurons that governs gut function — without surgery.
While optogenetics has been used for brain research since the early 2000s, this marks the first non-invasive platform for wireless optical stimulation of the gut, opening new possibilities for mapping neural circuits that were previously inaccessible to researchers.
ICOPS represents the latest in Ramadi's portfolio of ingestible technologies, which includes FLASH, a capsule that uses electrical stimulation to activate gut neurons, and IMAG, a magnetic field-based device for tracking pill location in the gut. While these other devices have shown that neural activation can lead to hormonal changes affecting metabolism, ICOPS adds optogenetic control for greater precision.
A key innovation is that ICOPS operates without a battery, instead receiving power wirelessly through magnetic induction from an external transmitter. This battery-free design was necessary for the device to be small enough for testing in rats.
"What makes this capsule unique is that it was entirely fabricated in-house using 3D printing, without the need for cleanroom facilities,” said Mohamed Elsherif, a Postdoctoral Associate in Ramadi’s lab and the paper’s lead author. “This allowed us to integrate micro-LEDs and custom coils in a scalable way, making it the first rodent-scale ingestible capsule for non-invasive optical stimulation. Crucially, it can operate wirelessly in freely moving animals, enabling studies that were not possible with traditional tethered or invasive approaches."
The implications extend beyond research. The technology could lead to new treatments for gut motility disorders. "We don't really have very good prokinetic or antikinetic agents," Ramadi said, referring to drugs that speed up or slow down gut movement. "We have stuff that overall slows or accelerates motility, but not targeted ones."
Neural activation in specific gut regions can also trigger hormonal changes affecting metabolism, potentially offering new approaches to treating metabolic diseases and eating disorders.
The devices travel through the digestive system naturally over one to two days. Beyond light therapy, the platform could enable electrical stimulation and targeted drug delivery. While clinical applications likely remain a decade away, the research represents a significant step toward understanding the gut's complex neural networks.
In addition to Ramadi and Elsherif, the paper's authors are Rawan Badr El-Din, Zhansaya Makhambetova, Heba Naser, Rahul Singh, Keonghwan Oh, and Revathi Sukesan from NYUAD's Division of Engineering; Maylis Boitet from NYUAD's Core Technology Platforms Operations; and Sohmyung Ha from NYUAD's Division of Engineering and NYU Tandon.
Funding for the ICOPS research came from NYUAD and Tamkeen under the NYUAD Research Institute Award to the Research Center for Translational Medical Devices (CENTMED).
Elsherif, Mohamed, et al. “Wirelessly powered ingestible capsule for optical stimulation of the gastrointestinal tract in rodents.” Advanced Materials Technologies, 20 Aug. 2025
Remote work spurs grassroots environmental action in New York City
Remote and hybrid work arrangements enabled New Yorkers to participate in community environmental action by giving them both the time and the motivation to do so, a new study from NYU finds.
Published in the Proceedings of the ACM on Human-Computer Interaction and presented at the 2025 Aarhus Conference on Critical Computing, the study draws on five years of ethnographic research at a volunteer-run composting and gardening site in Sunnyside, Queens.
It found that flexible schedules and work-from-home routines made it possible for independent and creative workers to engage in hands-on environmental labor during the workday. Many reported being driven not only by availability but by a desire to counter the isolation and screen fatigue associated with remote professional life.
Conducted by Margaret Jack, Industry Assistant Professor in NYU Tandon’s Department of Technology, Culture, and Society, the research focuses on a site known as 45th St Greenspace, established in 2020 on a formerly vacant lot.
Volunteers — many of whom were freelance or hybrid workers in fields like design, academia, or media — organized composting operations, garden plantings, public events, and infrastructure improvements. Most lived nearby and integrated the garden into their daily or weekly routines.
“Working from home didn’t just change where people did their jobs, it changed how they lived in their neighborhoods,” said Jack. “We found that flexible schedules and a need for offline connection drew people into environmental projects, where they could turn screen time into green time.”
The study offers new insight into how technology-mediated work environments shape civic participation and local infrastructure. It contributes to human-centered engineering research by examining how digital platforms like Slack, Zoom, and Signal also alter grassroots environmental systems and collective organizing.
Participants coordinated their work using these platforms, and the project’s governance structure reflected common patterns in digitally enabled work: horizontal decision-making, collaborative workflows, and distributed leadership.
Jack frames the garden itself as a socio-technical system, one whose function and sustainability depended not just on physical tools like compost bins and raised beds, but also on the digital infrastructure and cultural norms imported from participants’ professional lives.
The project reflects how technologies designed for individual productivity are being adapted for civic and ecological collaboration, raising design questions relevant to the development of future civic technologies.
While the project was open to the public and built on values of inclusion and mutual aid, Jack found that sustained participation was shaped by access to time, stability, and professional autonomy. Parents of young children, people with rigid or physically demanding jobs, and those unfamiliar with digital communication tools were often less able to remain actively involved.
Cultural expectations around communication and conflict resolution also revealed differences within the group. Though mediation structures were developed, they often relied on middle-class professional norms, which did not resonate equally across cultural or linguistic backgrounds.
The research employed a combination of ethnographic and autoethnographic methods. Jack, a local resident and volunteer at the garden, conducted participant observation over five years, maintained field notes and reflective journals, and led interviews and a survey with core volunteers.
The project also included contributions from community collaborators and comparative fieldwork in other New York gardens. These methods, often used in the design and evaluation of socio-technical systems, allow engineering researchers to understand not only how tools function, but how they shape behavior, governance, and access.
Though 45th St Greenspace is now preparing to close due to private development, Jack sees the project as emblematic of a wider shift in urban civic engagement. As hybrid and independent work becomes more common, she argues, it reshapes how people interact with both digital platforms and the physical city, bringing new possibilities for environmental infrastructure, but also new forms of exclusion.
Margaret Jack. 2025. In the Dirt: Place-Based Environmental Action and Technology-Mediated Work in New York City. In Proceedings of the sixth decennial Aarhus conference: Computing X Crisis (AAR '25). Association for Computing Machinery, New York, NY, USA, 96–106.
NYU Tandon researcher advocates for uncertainty-aware water risk models to improve flood and drought preparedness
Researchers are calling for a more reliable approach to understanding water-related hazards by explicitly accounting for uncertainty in their predictions, arguing this could improve how communities prepare for the risk of floods, droughts, and river-related erosion.
Omar Wani, a hydrologist at NYU Tandon School of Engineering, and co-authors argue in a recent opinion piece published in PLOS Water that many current hydroclimatic hazard assessments have a major flaw: they only give one answer. These models might predict, for example, that a river will flood to 15 feet, but they don't say how confident scientists are in that prediction or what other outcomes are possible.
Wani, who joined NYU Tandon as an Assistant Professor in the Civil and Urban Engineering Department in 2023, leads the Hydrologic Systems Group, which combines statistical and computational methods to study water dynamics in built and natural environments. His group focuses on understanding hydroclimatic risk and enabling more reliable decision-making under uncertainty.
This uncertainty-focused approach is central to Wani and his PLOS co-authors' argument for models that work more like weather forecasts, giving a range of possibilities with probabilities attached. Rather than saying "the water level in the river will reach 15 feet," these models might say "there's an 80% chance of the water level exceeding 15 feet, a 30% chance of it exceeding 18 feet, and a 10% chance of it breaching the 20 feet mark."
The approach has real-world urgency. Approximately 75% of flood-related fatalities occur when people drive into or walk through floodwaters, while climate change is expected to cause additional capacity deficits in stormwater infrastructure, leading to enormous financial losses. Overwhelmed and damaged drainage structures under roads can cost millions to replace.
In research published in Earth Surface Dynamics, Wani and collaborators demonstrated practical applications of this approach, showing how probabilistic models can generate "geomorphic risk maps" that display the probability of riverbank erosion at different locations over time.
Using satellite data from the rapidly-migrating Ucayali River in Peru's Amazon basin, the researchers showed their novel probabilistic approach consistently outperformed traditional predictions. The method combines mathematical models based on river shape and curves with computer simulations that run thousands of different scenarios to explore possible future outcomes.
Apart from the scientific value of this research in improving our understanding of the river systems, such "risk maps are relatively more informative in avoiding false negatives, which can be both detrimental and costly, in the context of assessing erosional hazards," said Wani. Their results showed that probabilistic forecasts assign appropriate probabilities to regions that might erode, avoiding the overconfident binary classifications of traditional approaches.
The implications extend beyond academic research. Behavioral science research shows that people can exhibit loss aversion and risk aversion when making decisions under uncertainty. However, these psychological preferences can only be utilized when the requisite uncertainty information is available.
"To allow for individuals to use these preferences and risk attitudes during hydroclimatic warning or design decisions, people would need to be aware of the uncertainties in quantitative analysis and forecasts," Wani explained.
His group's current work spans from improving the reliability of flood early warning systems for distributed stormwater infrastructure to testing advanced probabilistic algorithms for satellite-based flood damage classification.
The framework represents a shift from seeking the single "most likely" outcome to embracing the full range of possibilities.
The research has immediate practical applications for infrastructure planning, emergency management, and community resilience. As climate change introduces additional uncertainties into the behavior of streams and rivers globally, the researchers argue that probabilistic approaches become increasingly important. The work reflects growing recognition that uncertainty is not a limitation to overcome, but rather crucial information that enables better decision-making.
In addition to Wani, the PLOS opinion piece's authors are Mason Majszak, who is currently working on a Swiss National Science Foundation project as a Postdoctoral Fellow at NYU Tandon and in the NYU Department of Philosophy, Victor Hertel from the German Aerospace Center, and Christian Geiß from the German Aerospace Center and University of Bonn. Funding for the work was provided by the Swiss National Science Foundation.
The Earth Surface Dynamics paper's authors are, in addition to Wani, Brayden Noh from Caltech, Kieran B. J. Dunne from Caltech and Delft University of Technology, and Michael P. Lamb from Caltech. Funding for the research was provided by the Swiss National Science Foundation and the Resnick Sustainability Institute at Caltech under National Science Foundation awards.
Climate change is altering nitrogen composition in Arctic rivers, study finds
Climate change is starving the Arctic Ocean of essential nutrients, with the region's six largest rivers now delivering far less of the type of nitrogen that marine ecosystems need to survive, according to new research in one of Earth's most vulnerable regions.
The study, led by Bridger J. Ruyle of NYU Tandon School of Engineering, is published in Global Biogeochemical Cycles, where it has been selected as an Editor's Choice. Ruyle completed the research as a Postdoctoral Fellow at the Carnegie Institution for Science.
The study found that warming temperatures and thawing permafrost are fundamentally altering the chemistry of Arctic rivers. The result is that coastal food webs that have sustained Indigenous communities for millennia are being deprived of inorganic nitrogen, an essential nutrient, potentially triggering cascading effects throughout the Arctic Ocean ecosystem.
"This is a red flag for the Arctic," said Ruyle, who joined NYU Tandon in the summer of 2025 as an Assistant Professor in the Civil and Urban Engineering Department. "Rapid changes in river nitrogen chemistry could completely transform how these marine ecosystems function."
The research analyzed 20 years of data from six major Arctic rivers — the Yenisey, Lena, Ob', Mackenzie, Yukon, and Kolyma — which collectively drain two-thirds of the land area flowing into the Arctic Ocean. These rivers transport nitrogen that supports up to 66% of the ecosystem's primary production in coastal Arctic regions.
Between 2003 and 2023, Ruyle and colleagues documented declines in inorganic nitrogen accompanied by simultaneous increases in dissolved organic nitrogen, a far less bioavailable form of the element, in four of the six rivers. The findings reveal that warmer temperatures and increased precipitation caused by climate change are driving the shift in nitrogen composition through their effects on river discharge and permafrost thaw.
Using sophisticated statistical modeling, the researchers identified permafrost loss as the key factor explaining the diverging trends between organic and inorganic nitrogen in these rivers. The study combined 20 years of water chemistry data with environmental variables including temperature, precipitation, land cover, and permafrost extent to pinpoint the climate drivers behind the chemical shifts.
This Arctic rivers research represents Ruyle's broader research mission to understand how human activity, climate change, and natural processes interact to affect water quality globally. Among other areas of focus, his work includes tracking "forever chemicals" and pharmaceuticals in wastewater.
"Whether we're looking at PFAS contamination in drinking water or nitrogen cycling in Arctic rivers, the common thread is understanding how environmental changes propagate through water systems," Ruyle explained. His research explores how human activity, the biosphere, and climate change affect water quality, with particular focus on developing analytical tools to quantify chemical contamination and developing models using remote sensing data to assess climate impacts.
The Arctic findings have implications for ecosystem management and climate adaptation strategies. River transport of nitrogen is estimated to support up to 66% of primary production in Arctic coastal regions, making these compositional changes important for marine food webs and the Indigenous communities that depend on these resources.
The research also highlights the interconnected nature of global environmental challenges. As Ruyle noted in previous work on pharmaceutical contamination, climate-driven water scarcity could exacerbate water quality problems, as there's less dilution of contaminants during drought conditions. The Arctic study similarly shows how temperature and precipitation changes cascade through complex biogeochemical systems, resulting in water quality and ecosystem impacts
"This work demonstrates why we need to think about water quality and climate change as fundamentally linked challenges," Ruyle said. " As climate change intensifies, we must understand these interconnections to protect both human health and ecosystem integrity."
Along with Ruyle, the paper's authors are Julian Merder of the University of Canterbury, New Zealand; Robert G.M. Spencer of Florida State University; James W. McClelland of the Marine Biological Laboratory, Woods Hole; Suzanne E. Tank of the University of Alberta; and Anna M. Michalak of Carnegie Institution for Science.
The study was supported by the National Science Foundation through grants for the Arctic Great Rivers Observatory.
Ruyle, B. J., Merder, J., Spencer, R. G. M., McClelland, J. W., Tank, S. E., & Michalak, A. M. (2025). Changes in the composition of nitrogen yields in large Arctic rivers linked to temperature and precipitation. Global Biogeochemical Cycles, 39, e2025GB008639. https://doi.org/10.1029/2025GB008639
NYU Tandon engineers develop new transparent electrode for infrared cameras
Infrared imaging helps us see things the human eye cannot. The technology — which can make visible body heat, gas leaks or water content, even through smoke or darkness — is used in military surveillance, search and rescue missions, healthcare applications and even in autonomous vehicles.
These capabilities come with an engineering challenge, however. Infrared cameras need electrical contacts to capture and transmit the images they detect. Most materials that can conduct electrical signals also block the majority of infrared radiation from reaching the sensor, creating a fundamental conflict between seeing infrared light and having the electrical connections needed to process that information.
To solve this, researchers at NYU Tandon School of Engineering have developed a transparent electrode made from embedding tiny silver wires, similar in width to human hair, into a transparent plastic matrix that can be simply deposited on top of conventional infrared detectors.
The research, published in the Journal of Materials Chemistry C and recently selected as a HOT article by the journal's editors, tackles a key challenge in infrared detector manufacturing.
"We've developed a material that solves a fundamental problem that has been limiting infrared detector design," said Ayaskanta Sahu, associate professor in the Department of Chemical and Biomolecular Engineering (CBE) at NYU Tandon and the study's senior author. "Our transparent electrode material works well across the infrared spectrum, giving engineers more flexibility in how they build these devices."
The researchers tested their material by building it into infrared cameras that use colloidal quantum dots as the light-responsive material. These are tiny engineered particles that have recently gained attention for their use in quantum dot televisions and their role in earning the 2023 Nobel Prize in Chemistry. For this study, the group specifically used tiny clusters of mercury telluride, a type of quantum dot that responds to various wavelengths of infrared light.
Their new approach represents a significant improvement over existing methods. Traditional infrared photodetectors have relied on expensive materials like indium tin oxide (ITO) or thin metal films, which either lose transparency in longer infrared wavelengths or suffer from poor electrical properties and must be rigid.
Measuring 120 nanometers in diameter and 10-30 micrometers in length, the silver nanowires form conductive networks even at relatively low concentrations. When embedded in the PVA matrix, they form a silvery conductive ink that can be sprayed or spun onto infrared detectors as stable and flexible films that can even be manufactured at the low temperatures needed for quantum dot processing.
"Conventional electrodes in the infrared are like blackout curtains — most of the signal never reaches the sensor," said graduate researcher Shlok J. Paul, a co-author on the study. "Our near-invisible web of silver nanowires lets more infrared photons in while doubling as the wiring that carries the electrical current needed to turn the invisible light into data. While there is more work to be done, the simplicity of this flexible layer could carry IR detection from the lab to commercial applications like for firefighter vision or self-driving cars.”
In addition to Sahu and Paul, the paper's authors are Elisa Riedo, Herman F. Mark Professor in NYU Tandon's CBE Department, and graduate students Håvard Mølnås, Steven L. Farrell, and Nitika Parashar, all from CBE as well.
The work was supported by the Defense Advanced Research Projects Agency, the Office of Naval Research, the US Army Research Office, and the National Science Foundation.
The researchers filed a U.S. patent application covering their method for embedding silver nanowires in a polymer matrix for transparent infrared electrodes.
Paul, Shlok J., et al. “Plenty of room at the top: Exploiting nanowire – polymer synergies in transparent electrodes for infrared imagers.” Journal of Materials Chemistry C, vol. 13, no. 21, 2025, pp. 10592–10601
NYU Tandon researchers develop AI agent that solves cybersecurity challenges autonomously
Artificial intelligence agents — AI systems that can work independently toward specific goals without constant human guidance — have demonstrated strong capabilities in software development and web navigation. Their effectiveness in cybersecurity has remained limited, however.
That may soon change, thanks to a research team from NYU Tandon School of Engineering, NYU Abu Dhabi and other universities that developed an AI agent capable of autonomously solving complex cybersecurity challenges.
The system, called EnIGMA, was presented this month at the International Conference on Machine Learning (ICML) 2025 in Vancouver, Canada.
"EnIGMA is about using Large Language Model agents for cybersecurity applications," said Meet Udeshi, a NYU Tandon Ph.D. student and co-author of the research. Udeshi is advised by Ramesh Karri, Chair of NYU Tandon's Electrical and Computer Engineering Department (ECE) and a faculty member of the NYU Center for Cybersecurity and NYU Center for Advanced Technology in Telecommunications (CATT), and by Farshad Khorrami, ECE professor and CATT faculty member. Both Karri and Khorrami are co-authors on the paper, with Karri serving as a senior author.
To build EnIGMA, the researchers started with an existing framework called SWE-agent, which was originally designed for software engineering tasks. However, cybersecurity challenges required specialized tools that didn't exist in previous AI systems. "We have to restructure those interfaces to feed it into an LLM properly. So we've done that for a couple of cybersecurity tools," Udeshi explained.
The key innovation was developing what they call "Interactive Agent Tools" that convert visual cybersecurity programs into text-based formats the AI can understand. Traditional cybersecurity tools like debuggers and network analyzers use graphical interfaces with clickable buttons, visual displays, and interactive elements that humans can see and manipulate.
"Large language models process text only, but these interactive tools with graphical user interfaces work differently, so we had to restructure those interfaces to work with LLMs," Udeshi said.
The team built their own dataset by collecting and structuring Capture The Flag (CTF) challenges specifically for large language models. These gamified cybersecurity competitions simulate real-world vulnerabilities and have traditionally been used to train human cybersecurity professionals.
"CTFs are like a gamified version of cybersecurity used in academic competitions. They're not true cybersecurity problems that you would face in the real world, but they are very good simulations," Udeshi noted.
Paper co-author Minghao Shao, a NYU Tandon Ph.D. student and Global Ph.D. Fellow at NYU Abu Dhabi who is advised by Karri and Muhammad Shafique, Professor of Computer Engineering at NYU Abu Dhabi and ECE Global Network Professor at NYU Tandon, described the technical architecture: "We built our own CTF benchmark dataset and created a specialized data loading system to feed these challenges into the model." Shafique is also a co-author on the paper.
The framework includes specialized prompts that provide the model with instructions tailored to cybersecurity scenarios.
EnIGMA demonstrated superior performance across multiple benchmarks. The system was tested on 390 CTF challenges across four different benchmarks, achieving state-of-the-art results and solving more than three times as many challenges as previous AI agents.
During the research conducted approximately 12 months ago, "Claude 3.5 Sonnet from Anthropic was the best model, and GPT-4o was second at that time," according to Udeshi.
The research also identified a previously unknown phenomenon called "soliloquizing," where the AI model generates hallucinated observations without actually interacting with the environment, a discovery that could have important consequences for AI safety and reliability.
Beyond this technical finding, the potential applications extend outside of academic competitions. "If you think of an autonomous LLM agent that can solve these CTFs, that agent has substantial cybersecurity skills that you can use for other cybersecurity tasks as well," Udeshi explained. The agent could potentially be applied to real-world vulnerability assessment, with the ability to "try hundreds of different approaches" autonomously.
The researchers acknowledge the dual-use nature of their technology. While EnIGMA could help security professionals identify and patch vulnerabilities more efficiently, it could also potentially be misused for malicious purposes. The team has notified representatives from major AI companies including Meta, Anthropic, and OpenAI about their results.
In addition to Karri, Khorrami, Shafique, Udeshi and Shao, the paper's authors are Talor Abramovich (Tel Aviv University), Kilian Lieret (Princeton University), Haoran Xi (NYU Tandon), Kimberly Milner (NYU Tandon), Sofija Jancheska (NYU Tandon), John Yang (Stanford University), Carlos E. Jimenez (Princeton University), Prashanth Krishnamurthy (NYU Tandon), Brendan Dolan-Gavitt (NYU Tandon), Karthik Narasimhan (Princeton University), and Ofir Press (Princeton University).
Funding for the research came from Open Philanthropy, Oracle, the National Science Foundation, the Army Research Office, the Department of Energy, and NYU Abu Dhabi Center for Cybersecurity and Center for Artificial Intelligence and Robotics.