New FEMA head may force agency to leave DC because he 'refuses to leave Texas': report

Nicole Charky-Chami
November 17, 2025  

RAW STORY

Photo credit: (Texas Division of Emergency Management)

A top emergency official is reportedly refusing to leave Texas — and potentially could force the Federal Emergency Management Agency to move to the Lone Star State.

This could present "huge challenges" for the agency to coordinate with the Department of Homeland Security, a former FEMA official told Politico on Monday.

DHS oversees FEMA, whose top leader announced he was stepping down from the top role, and his potential replacement — Nim Kidd, head of the Texas Division of Emergency Management — could be up for the job next.

The move follows the news that David Richardson, a Kristi Noem ally who was criticized after he was publicly silent for a week following floods in Texas that killed more than 130 people, announced that he plans to resign. Department of Homeland Security officials reportedly had planned to remove the FEMA chief from the role after his six-month stint.

FEMA chief of staff Karen Evans will take on the interim position of administrator beginning Dec. 1.

A former FEMA official told The Politico that a panel appointed by President Donald Trump is planning to consider moving the agency to Texas to oblige Kidd, who apparently did not want to leave his home state, and is one of the 13 members of the review panel who remains close to the Trump administration.

“The admin wanted him, but he refused to leave Texas,” a FEMA source told the outlet.

 

Teaching large language models how to absorb new knowledge

With a new method developed at MIT, an LLM behaves more like a student, writing notes that it studies to memorize new information.

Massachusetts Institute of Technology

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

CAMBRIDGE, MA -- In an MIT classroom, a professor lectures while students diligently write down notes they will reread later to study and internalize key information ahead of an exam.

Humans know how to learn new information, but large language models can’t do this in the same way. Once a fully trained LLM has been deployed, its “brain” is static and can’t permanently adapt itself to new knowledge.

This means that if a user tells an LLM something important today, it won’t remember that information the next time this person starts a new conversation with the chatbot.

Now, a new approach developed by MIT researchers enables LLMs to update themselves in a way that permanently internalizes new information. Just like a student, the LLM generates its own study sheets from a user’s input, which it uses to memorize the information by updating its inner workings.

The model generates multiple self-edits to learn from one input, then applies each one to see which improves its performance the most. This trial-and-error process teaches the model the best way to train itself.

The researchers found this approach improved the accuracy of LLMs at question-answering and pattern-recognition tasks, and it enabled a small model to outperform much larger LLMs.

While there are still limitations that must be overcome, the technique could someday help artificial intelligence agents consistently adapt to new tasks and achieve changing goals in evolving environments.    

“Just like humans, complex AI systems can’t remain static for their entire lifetimes. These LLMs are not deployed in static environments. They are constantly facing new inputs from users. We want to make a model that is a bit more human-like — one that can keep improving itself,” says Jyothish Pari, an MIT graduate student and co-lead author of a paper on this technique.

He is joined on the paper by co-lead author Adam Zweiger, an MIT undergraduate; graduate students Han Guo and Ekin Akyürek; and senior authors Yoon Kim, an assistant professor in the Department of Electrical Engineering and Computer Science (EECS) and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL), and Pulkit Agrawal, an assistant professor in EECS and member of CSAIL. The research will be presented at the Conference on Neural Information Processing Systems.

Teaching the model to learn

LLMs are neural network models that have billions of parameters, called weights, that contain the model’s knowledge and process inputs to make predictions. During training, the model adapts these weights to learn new information contained in its training data.

But once it is deployed, the weights are static and can’t be permanently updated anymore.

However, LLMs are very good at a process called in-context learning, in which a trained model learns a new task by seeing a few examples. These examples guide the model’s responses, but the knowledge disappears before the next conversation.

The MIT researchers wanted to leverage a model’s powerful in-context learning capabilities to teach it how to permanently update its weights when it encounters new knowledge.

The framework they developed, called SEAL for “self-adapting LLMs,” enables an LLM to generate new synthetic data based on an input, and then determine the best way to adapt itself and learn from that synthetic data. Each piece of synthetic data is a self-edit the model can apply.

In the case of language, the LLM creates synthetic data by rewriting the information, and its implications, in an input passage. This is similar to how students make study sheets by rewriting and summarizing original lecture content.

The LLM does this multiple times, then quizzes itself on each self-edit to see which led to the biggest boost in performance on a downstream task like question answering. It uses a trial-and-error method known as reinforcement learning, where it receives a reward for the greatest performance boost.

Then the model memorizes the best study sheet by updating its weights to internalize the information in that self-edit.

“Our hope is that the model will learn to make the best kind of study sheet — one that is the right length and has the proper diversity of information — such that updating the model based on it leads to a better model,” Zweiger explains.

Choosing the best method

Their framework also allows the model to choose the way it wants to learn the information. For instance, the model can select the synthetic data it wants to use, the rate at which it learns, and how many iterations it wants to train on.

In this case, not only does the model generate its own training data, but it also configures the optimization that applies that self-edit to its weights.

“As humans, we know how we learn best. We want to grant that same ability to large language models. By providing the model with the ability to control how it digests this information, it can figure out the best way to parse all the data that are coming in,” Pari says.

SEAL outperformed several baseline methods across a range of tasks, including learning a new skill from a few examples and incorporating knowledge from a text passage. On question answering, SEAL improved model accuracy by nearly 15 percent and on some skill-learning tasks, it boosted the success rate by more than 50 percent.

But one limitation of this approach is a problem called catastrophic forgetting: As the model repeatedly adapts to new information, its performance on earlier tasks slowly declines.

The researchers plan to mitigate catastrophic forgetting in future work. They also want to apply this technique in a multi-agent setting where several LLMs train each other.

“One of the key barriers to LLMs that can do meaningful scientific research is their inability to update themselves based on their interactions with new information. Though fully deployed self-adapting models are still far off, we hope systems able to learn this way could eventually overcome this and help advance science,” Zweiger says.

###

This work is supported, in part, by the U.S. Army Research Office, the U.S. Air Force AI Accelerator, the Stevens Fund for MIT UROP, and the MIT-IBM Watson AI Lab.

---

DOI

10.48550/arXiv.2506.10943 

 

Student innovation connects wildfire resilience, safety to home design

UBC Okanagan Engineering students create a guide to sustainable, fire-resilient housing suited to BC’s interior

University of British Columbia Okanagan campus

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

A rendering of EcoHaven, a student-designed modular, wildfire-resilient and net-zero home concept created by UBC Okanagan and Thompson Rivers University students for the 2024 U.S. Department of Energy Solar Decathlon.

view more 

Credit: UBC Okanaga

Two UBC Okanagan engineering students are transforming classroom research into a practical tool for communities facing increasing wildfire risk. 

Under the supervision of Dr. Qian Chen, Miracle Kabano and Samantha Krieg co-authored a new paper outlining the Wildfire-resilient and Sustainable Evaluation Framework for British Columbia (WiSE-BC).  

The study appears in Lecture Notes in Civil Engineering and builds directly on the students’ earlier success designing EcoHaven, a modular home that won international recognition for wildfire resilience and energy efficiency. 

The EcoHaven project—developed in collaboration with Thompson Rivers University faculty Dr. Dale Parkes and Dr. Hossein (Sayed) Banitabaei, along with a multidisciplinary student team and industry partners—earned second place in the US Department of Energy’s 2024 Solar Decathlon Design Challenge.  

Designed for Honour Ranch, a retreat near Ashcroft, BC, that supports veterans and first responders, EcoHaven combines wildfire-resistant materials, net-zero energy systems and affordability suited to BC communities. 

When Dr. Chen and her students later developed WiSE-BC, they used EcoHaven as a test case to evaluate the framework’s real-world potential.  

WiSE-BC applies the analytical hierarchy process, a structured decision-making method that allows scalability and adaptability depending on project size and stakeholder priorities. This makes it suitable for both single-family builds and community-scale planning. 

The results showed that WiSE-BC can help builders and designers identify trade-offs early, balancing emissions, cost and resilience at the concept stage.  

In practical terms, that means reducing design time and construction costs while improving sustainability and fire-safety outcomes. 

“With WiSE-BC, we wanted to explore and bring attention to an industry gap of both wildfire resilience and sustainability in design,” says Kabano. “Presenting our research at the Canadian Society for Civil Engineering conference was an incredible opportunity to help BC communities and developers make better design decisions in the early stages of a project.” 

“British Columbia urgently needs housing that can withstand climate extremes,” adds Dr. Chen, Assistant Professor of Civil Engineering. “WiSE-BC provides a roadmap for sustainable design that can be adopted by builders today, not years from now.” 

Krieg says leading the EcoHaven project and co-authoring WiSE-BC revealed how student-driven collaboration can have lasting changes. 

“It showed me the material impact that students can have on the world when they work together and strive for something greater,” Krieg says. “By translating that work into research publications that offer practical solutions for industry, we hope to inspire others to build better in BC.” 

She adds that the experience shaped her career ambitions. 

“It inspired me to pursue a doctorate and continue investigating the intersection of sustainability and disaster resilience,” she says.  

The same student research group is now developing two additional papers based on the EcoHaven design and a related project from the previous year. As housing demand and wildfire threats continue to rise, the team hopes WiSE-BC and its successors will guide municipalities, homebuilders and policymakers toward practical, evidence-based design solutions. 

 

---

DOI

10.2139/ssrn.5647232 

Method of Research

Computational simulation/modeling

Subject of Research

Not applicable

Article Title

Decision Framework for the Evaluation of Wildfire Resilience and Sustainability in Single-Family Housing of British Columbia Interior (WiSE-BC)

 Chronic Wasting Disease

Almost half of Oregon elk population carries advantageous genetic variant against CWD, study shows

University of Illinois College of Agricultural, Consumer and Environmental Sciences

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

 

image: 

A Roosevelt elk at the Dean Creek Elk Viewing Area near Reedsport, Oregon. 

view more 

Credit: Martyne Reesman, Oregon Department of Fish and Wildlife

URBANA, Ill. – Chronic Wasting Disease, a prion protein disease that is fatal in elk, deer, and other cervids, has spread rapidly across the United States since it was first identified in 1967. CWD has now reached Idaho near the Oregon border, causing concern for the Columbian white-tailed deer, a rare subspecies found only in two regions in Oregon.

The deer have little genetic protection against CWD, but a new study from the University of Illinois Urbana-Champaign shows that about half of Oregon elk carry a gene that makes them less susceptible to the disease. That could help slow down CWD’s progress into the state, but it is still important to monitor and manage cervid populations.

“CWD has not been detected in Oregon yet, but it’s almost inevitable that it will happen. We are collaborating with colleagues in Oregon to examine how quickly CWD is likely to spread once it gets into the state. We are studying how common the advantageous variant of the prion gene is in Oregon’s elk population,” said lead and co-corresponding author Yasuko Ishida, senior research specialist in the Department of Animal Sciences, part of the College of Agricultural, Consumer and Environmental Sciences at U. of I.

CWD is caused by abnormally folded proteins known as prions, and it is related to scrapie in sheep and Creutzfeldt-Jakob disease in humans. The disease leads to progressive neurological damage, and it is always fatal. CWD is spread both by direct contact and through the environment, and it readily crosses between cervid species such as elk, deer, and mule deer.

While normal prion proteins exist in all mammals, they can misfold when an infectious prion enters the host. That initiates a chain reaction causing normal prion proteins to change shape and go rogue. Past research has shown that some variants of the gene for prion protein synthesis (PRNP) are associated with lower susceptibility to the disease. For elk, a codon that encodes the amino acid leucine rather than methionine is more frequently found in CWD-negative than -positive elk, indicating a genetic advantage for those animals. 

“We previously tested a small number of Columbian white-tailed deer in Oregon’s Douglas County, and we found that they were genetically very disadvantaged when it came to CWD. We didn’t find the protective variant of the gene in a single deer,” said corresponding author Alfred Roca, professor of animal sciences at Illinois.

“That’s why we wanted to look at the Oregon elk population, because if the elk get CWD, it's quite likely that Columbian white-tailed deer will get it as well.”

The researchers sequenced the coding region of PRNP in 183 elk collected across Oregon, including 82 Rocky Mountain elk from the eastern part of the state, and 101 Roosevelt elk found west of the Cascade Mountains. This is the first study to genotype the Roosevelt elk for the prion gene, Roca noted.

Ishida and Roca found that 42% of Roosevelt elk and 49% of Rocky Mountain elk carried at least one copy of the allele that encodes for leucine. That is a relatively high proportion compared to elk populations nationwide, they said.

“Without advantageous genetic variants, a whole population can be wiped out fairly quickly. It’s good to know that at least some of the elk in Oregon will be less susceptible. It’s not enough to prevent the spread of CWD once it enters the state, but it may slow it down a bit,” Roca said.

The Oregon Department of Fish and Wildlife (ODFW) tests extensively for CWD in elk and deer populations. Anyone who hunts and harvests an animal or wants to claim a vehicle-struck carcass can submit the head to a testing facility.

If an infection were detected, the ODFW would work to reduce herd density, said Colin Gillin, ODWF state wildlife veterinarian and a co-author on the study.

“Depending on the species, we may reduce herds in affected areas to keep the nose-to-nose contact at a bare minimum. There's absolutely no evidence of CWD crossing into humans, but people may still be reluctant to hunt if the disease shows up in their area. If we can keep the disease at low prevalence and prevent it from spreading across the 98,000 square miles of the state, it's a big benefit biologically, socially, and economically.”

“These susceptibility studies we conduct with the University of Illinois are hugely important; they inform the timing and type of management actions we would use,” he said.

The paper, “PRNP variant frequencies in Roosevelt and Rocky Mountain elk (Cervus canadensis) from Oregon and their implications for chronic wasting disease,” is published in the Journal of Heredity [DOI: 10.1093/jhered/esaf096]. The funding was provided by the USDA’s Animal and Plant Health Inspection Service.

---

Journal

Journal of Heredity

DOI

10.1093/jhered/esaf096 

Article Title

PRNP variant frequencies in Roosevelt and Rocky Mountain elk (Cervus canadensis) from Oregon and their implications for chronic wasting disease

Article Publication Date

17-Nov-2025

 

Eating lean beef as part of a healthy diet may not increase heart disease risk

Moderate portions of lean beef as part of a Mediterranean diet did not increase one indicator of heart disease risk, according to a new study

Penn State

FacebookXLinkedInWeChatBlueskyMessageWhatsAppEmail

UNIVERSITY PARK, Pa. — Eating moderate amounts of lean beef as part of a Mediterranean diet does not increase an emerging risk factor for cardiovascular disease, according to a new study by an interdisciplinary research team at Penn State. The researchers examined indicators of heart health and gut microbiome diversity among relatively young and healthy participants who ate four different diets, including varying amounts and types of beef, for four weeks. 

In the study, recently published in Journal of the American Heart Association, the researchers found that eating a Mediterranean diet with either .5 or 2.5 ounces of lean beef each day did not increase blood levels of trimethylamine N-oxide (TMAO), an emerging indicator of risk for cardiovascular disease. This was in comparison to an average American diet with 2.5 ounces of regular beef each day and a third Mediterranean diet that included 5.5 ounces of lean beef each day. TMAO, a byproduct of metabolism, can be produced when people consume animal products including beef. 

“Observational evidence shows higher levels of TMAO are associated with higher cardiovascular risk,” said Kristina Petersen, associate professor of nutritional sciences and senior author of this study. “In this study we wanted to better understand the relationship between lean beef consumption and TMAO levels in the context of a healthy, Mediterranean style diet.” 

People have long been warned to limit beef intake because higher beef consumption has been linked to risk of heart disease, according to Petersen. These findings align with prior research examining other risk factors and suggest that moderate consumption of lean, unprocessed beef as part of a healthy diet does not worsen risk factors for heart disease. 

Measuring the impact of different diets 

The researchers analyzed samples from a previous study involving Petersen that examined beef in a Mediterranean diet.  For this analysis, 30 participants were provided with all meals and snacks for four separate four-week periods. In this experimental design, every participant consumed all four diets, which reduced the possibility that results could occur due to differences between the people eating each diet.

During one of the four-week study periods, participants consumed an average American diet, based on U.S data from the time the study was designed. In the American diet period, participants consumed meals that were composed of 52% carbohydrates, 15% proteins and 33% fats. Each day, they ate 2.5 ounces — about the size of a deck of cards — of regular beef, which contains more than 10% fat. The American diet is higher in saturated fats and lower in olive oil, fruits and vegetables than the Mediterranean diet, Petersen said. 

During the other periods of the study, participants consumed Mediterranean style diets that were composed of 42% carbohydrates, 17% proteins and 41% fats. The Mediterranean diets included more olive oil, fruits and vegetables than the American diet. 

During one of the Mediterranean diet periods, participants ate .5 ounces of beef each day, which is a cube of meat smaller than one inch and reflects the amount of red meat consumed in a traditional Mediterranean diet. During another Mediterranean diet period, participants ate 2.5 ounces of beef each day. During the other Mediterranean period, participants ate 5.5 ounces of beef each day. 

The beef consumed in the Mediterranean diet periods was either lean — less than 10% fat — or extra lean — less than 5% fat — while the beef in the American diet period was not lean. All beef in the study was unprocessed. 

Participants ate each diet in a random order and were given at least one week off between dietary periods. Researchers used three types of biological samples — blood, feces and urine — from each participant to measure TMAO levels and the diversity of participants’ gut microbiomes. 

Can lean beef be part of a healthy diet? 

When participants ate .5 or 2.5 ounces of lean beef as part of a Mediterranean diet, they had lower blood levels of TMAO than when they ate the American diet, according to the results of the study. When participants ate the American diet with 2.5 ounces of non-lean beef daily or the Mediterranean diet with 5.5 ounces of lean beef daily, their TMAO levels were not different. These findings suggest dietary quality was more important than the amount of beef eaten, the researchers said.

"We chose 2.5 ounces of lean beef because that approximates the amount of beef that the average American consumes each day,” said Zachary DiMattia, doctoral candidate in nutritional sciences and lead author of this study. “This study suggests that, in the context of a healthy dietary pattern, people may be able to include similar amounts of lean beef without increasing their TMAO levels. If people eat reasonable portions of lean, unprocessed beef as part of a Mediterranean-style diet, we would not expect this specific marker of cardiovascular disease risk to rise.”

In addition to TMAO levels, the researchers examined how the different diets affected the diversity of participants’ gut microbiomes. Results indicated that all three Mediterranean diets increased gut microbiome diversity compared to the American diet.  

The researchers said that further research is needed to understand the role the gut microbiome plays in the relationship between diet and TMAO levels. They agreed, however, that this study has implications for individuals who want to eat a healthy diet.

“Lean, moderately sized, unprocessed cuts of beef can be included as part of a healthy diet when people are consuming plenty of fruits, vegetables and healthy fats like olive oil,” DiMattia said.

More dietary impacts of lean beef 

This study focused on TMAO levels, but Petersen’s laboratory group previously explored other health effects of adding lean beef to a Mediterranean diet.

In a study from earlier this year using the same data, the researchers examined how beef consumption affected the blood vessel health. The researchers found that a Mediterranean diet with lean beef resulted in lower blood pressure than when participants consumed an American diet.

Additionally, doctoral student Fatemeh Jafari led a review of previous studies that examined whether red meat consumption raised TMAO levels. The literature review highlighted the complicated nature of TMAO, Petersen said. Just under half of the studies found that red meat increased TMAO, while the rest showed no increase in TMAO associated with beef consumption.

Healthy eating is essential 

The most important way to reduce risk, according to the researchers, is to establish healthy eating habits. They said that by consuming more vegetables, fruits and whole grains and reducing saturated fats, people can reduce their risk of heart disease. They also cautioned against taking these results out of context.

“This evidence does not mean you can necessarily eat a week’s worth of beef — for example, a single, 17.5-ounce steak — at one time and see the same results,” Petersen said. “Additionally, this recommendation does not extend to non-lean beef or processed meats like sausage or salami. Finally, these studies were conducted in relatively young, healthy individuals, so further research is needed in older people or anyone with elevated heart disease risk.”

Jingcheng Zhao, postdoctoral researcher in nutritional sciences; Fuhua Hao, assistant research professor of veterinary and biomedical sciences ; Sergei Koshkin, assistant research professor at the Penn State Huck Institute of Life Sciences; Jordan Bisanz, assistant professor of biochemistry and molecular biology and Dorothy Foehr and J. Lloyd Huck Early Career Chair in Host-Microbiome Interactions; Andrew Patterson, professor of molecular toxicology and of biochemistry and molecular biology; Jennifer Fleming, associate teaching professor of nutritional sciences; and Penny Kris Etherton, retired Evan Pugh University Professor of Nutritional Sciences, contributed to this research.

This research was funded by the National Cattlemen’s Beef Association — a contractor to the Beef Checkoff — and Penn State.

---

Journal

Journal of the American Heart Association

DOI

10.1161/JAHA.125.041063 

Method of Research

Randomized controlled/clinical trial

Subject of Research

People

Article Title

Effect of Varying Quantities of Lean Beef as Part of a Mediterranean‐Style Dietary Pattern on Gut Microbiota and Plasma, Fecal, and Urinary Metabolites: A Randomized Crossover Controlled Feeding Trial