Nature Climate Change, Published online: 29 April 2025; doi:10.1038/s41558-025-02328-8

It has been argued that parts of the climate system can experience rapid changes and that such tipping can be anticipated by early warning signals. Here the authors discuss the limitations of such indicators and common pitfalls in their application.

Today the U.S. House of Representatives passed the TAKE IT DOWN Act, giving the powerful a dangerous new route to manipulate platforms into removing lawful speech that they simply don't like. President Trump himself has said that he would use the law to censor his critics. The bill passed the Senate in February, and it now heads to the president's desk. 

The takedown provision in TAKE IT DOWN applies to a much broader category of content—potentially any images involving intimate or sexual content—than the narrower NCII definitions found elsewhere in the bill. The takedown provision also lacks critical safeguards against frivolous or bad-faith takedown requests. Services will rely on automated filters, which are infamously blunt tools. They frequently flag legal content, from fair-use commentary to news reporting. The law’s tight time frame requires that apps and websites remove speech within 48 hours, rarely enough time to verify whether the speech is actually illegal. As a result, online service providers, particularly smaller ones, will likely choose to avoid the onerous legal risk by simply depublishing the speech rather than even attempting to verify it.

Congress is using the wrong approach to helping people whose intimate images are shared without their consent. TAKE IT DOWN pressures platforms to actively monitor speech, including speech that is presently encrypted. The law thus presents a huge threat to security and privacy online. While the bill is meant to address a serious problem, good intentions alone are not enough to make good policy. Lawmakers should be strengthening and enforcing existing legal protections for victims, rather than inventing new takedown regimes that are ripe for abuse. 

The speed with which new technologies hit the market is nothing compared to the speed with which talented researchers find creative ways to use them, train them, even turn them into things we can’t live without. One such researcher is MIT MAD Fellow Alexander Htet Kyaw, a graduate student pursuing dual master’s degrees in architectural studies in computation and in electrical engineering and computer science.

Kyaw takes technologies like artificial intelligence, augmented reality, and robotics, and combines them with gesture, speech, and object recognition to create human-AI workflows that have the potential to interact with our built environment, change how we shop, design complex structures, and make physical things.

One of his latest innovations is Curator AI, for which he and his MIT graduate student partners took first prize — $26,000 in OpenAI products and cash — at the MIT AI Conference’s AI Build: Generative Voice AI Solutions, a weeklong hackathon at MIT with final presentations held last fall in New York City. Working with Kyaw were Richa Gupta (architecture) and Bradley Bunch, Nidhish Sagar, and Michael Won — all from the MIT Department of Electrical Engineering and Computer Science (EECS).

Curator AI is designed to streamline online furniture shopping by providing context-aware product recommendations using AI and AR. The platform uses AR to take the dimensions of a room with locations of windows, doors, and existing furniture. Users can then speak to the software to describe what new furnishings they want, and the system will use a vision-language AI model to search for and display various options that match both the user’s prompts and the room’s visual characteristics.

“Shoppers can choose from the suggested options, visualize products in AR, and use natural language to ask for modifications to the search, making the furniture selection process more intuitive, efficient, and personalized,” Kyaw says. “The problem we’re trying to solve is that most people don’t know where to start when furnishing a room, so we developed Curator AI to provide smart, contextual recommendations based on what your room looks like.” Although Curator AI was developed for furniture shopping, it could be expanded for use in other markets.

Another example of Kyaw’s work is Estimate, a product that he and three other graduate students created during the MIT Sloan Product Tech Conference’s hackathon in March 2024. The focus of that competition was to help small businesses; Kyaw and team decided to base their work on a painting company in Cambridge that employs 10 people. Estimate uses AR and an object-recognition AI technology to take the exact measurements of a room and generate a detailed cost estimate for a renovation and/or paint job. It also leverages generative AI to display images of the room or rooms as they might look like after painting or renovating, and generates an invoice once the project is complete.

The team won that hackathon and $5,000 in cash. Kyaw’s teammates were Guillaume Allegre, May Khine, and Anna Mathy, all of whom graduated from MIT in 2024 with master’s degrees in business analytics.

In April, Kyaw will give a TedX talk at his alma mater, Cornell University, in which he’ll describe Curator AI, Estimate, and other projects that use AI, AR, and robotics to design and build things.

One of these projects is Unlog, for which Kyaw connected AR with gesture recognition to build a software that takes input from the touch of a fingertip on the surface of a material, or even in the air, to map the dimensions of building components. That’s how Unlog — a towering art sculpture made from ash logs that stands on the Cornell campus — came about.

Unlog represents the possibility that structures can be built directly from a whole log, rather than having the log travel to a lumber mill to be turned into planks or two-by-fours, then shipped to a wholesaler or retailer. It’s a good representation of Kyaw’s desire to use building materials in a more sustainable way. A paper on this work, “Gestural Recognition for Feedback-Based Mixed Reality Fabrication a Case Study of the UnLog Tower,” was published by Kyaw, Leslie Lok, Lawson Spencer, and Sasa Zivkovic in the Proceedings of the 5th International Conference on Computational Design and Robotic Fabrication, January 2024.

Another system Kyaw developed integrates physics simulation, gesture recognition, and AR to design active bending structures built with bamboo poles. Gesture recognition allows users to manipulate digital bamboo modules in AR, and the physics simulation is integrated to visualize how the bamboo bends and where to attach the bamboo poles in ways that create a stable structure. This work appeared in the Proceedings of the 41st Education and Research in Computer Aided Architectural Design in Europe, August 2023, as “Active Bending in Physics-Based Mixed Reality: The Design and Fabrication of a Reconfigurable Modular Bamboo System.”

Kyaw pitched a similar idea using bamboo modules to create deployable structures last year to MITdesignX, an MIT MAD program that selects promising startups and provides coaching and funding to launch them. Kyaw has since founded BendShelters to build the prefabricated, modular bamboo shelters and community spaces for refugees and displaced persons in Myanmar, his home country.

“Where I grew up, in Myanmar, I’ve seen a lot of day-to-day effects of climate change and extreme poverty,” Kyaw says. “There’s a huge refugee crisis in the country, and I want to think about how I can contribute back to my community.”

His work with BendShelters has been recognized by MIT Sandbox, PKG Social Innovation Challenge, and the Amazon Robotics’ Prize for Social Good.

At MIT, Kyaw is collaborating with Professor Neil Gershenfeld, director of the Center for Bits and Atoms, and PhD student Miana Smith to use speech recognition, 3D generative AI, and robotic arms to create a workflow that can build objects in an accessible, on-demand, and sustainable way. Kyaw holds bachelor’s degrees in architecture and computer science from Cornell. Last year, he was awarded an SJA Fellowship from the Steve Jobs Archive, which provides funding for projects at the intersection of technology and the arts. 

“I enjoy exploring different kinds of technologies to design and make things,” Kyaw says. “Being part of MAD has made me think about how all my work connects, and helped clarify my intentions. My research vision is to design and develop systems and products that enable natural interactions between humans, machines, and the world around us.” 

As demand grows for more powerful and efficient microelectronics systems, industry is turning to 3D integration — stacking chips on top of each other. This vertically layered architecture could allow high-performance processors, like those used for artificial intelligence, to be packaged closely with other highly specialized chips for communication or imaging. But technologists everywhere face a major challenge: how to prevent these stacks from overheating.

Now, MIT Lincoln Laboratory has developed a specialized chip to test and validate cooling solutions for packaged chip stacks. The chip dissipates extremely high power, mimicking high-performance logic chips, to generate heat through the silicon layer and in localized hot spots. Then, as cooling technologies are applied to the packaged stack, the chip measures temperature changes. When sandwiched in a stack, the chip will allow researchers to study how heat moves through stack layers and benchmark progress in keeping them cool. 

"If you have just a single chip, you can cool it from above or below. But if you start stacking several chips on top of each other, the heat has nowhere to escape. No cooling methods exist today that allow industry to stack multiples of these really high-performance chips," says Chenson Chen, who led the development of the chip with Ryan Keech, both of the laboratory’s Advanced Materials and Microsystems Group.

The benchmarking chip is now being used at HRL Laboratories, a research and development company co-owned by Boeing and General Motors, as they develop cooling systems for 3D heterogenous integrated (3DHI) systems. Heterogenous integration refers to the stacking of silicon chips with non-silicon chips, such as III-V semiconductors used in radio-frequency (RF) systems.   

"RF components can get very hot and run at very high powers — it adds an extra layer of complexity to 3D integration, which is why having this testing capability is so needed," Keech says.

The Defense Advanced Research Projects Agency (DARPA) funded the laboratory's development of the benchmarking chip to support the HRL program. All of this research stems from DARPA's Miniature Integrated Thermal Management Systems for 3D Heterogeneous Integration (Minitherms3D) program.

For the Department of Defense, 3DHI opens new opportunities for critical systems. For example, 3DHI could increase the range of radar and communication systems, enable the integration of advanced sensors on small platforms such as uncrewed aerial vehicles, or allow artificial intelligence data to be processed directly in fielded systems instead of remote data centers.

The test chip was developed through collaboration between circuit designers, electrical testing experts, and technicians in the laboratory's Microelectronics Laboratory. 

The chip serves two functions: generating heat and sensing temperature. To generate heat, the team designed circuits that could operate at very high power densities, in the kilowatts-per-square-centimeter range, comparable to the projected power demands of high-performance chips today and into the future. They also replicated the layout of circuits in those chips, allowing the test chip to serve as a realistic stand-in. 

"We adapted our existing silicon technology to essentially design chip-scale heaters," says Chen, who brings years of complex integration and chip design experience to the program. In the 2000s, he helped the laboratory pioneer the fabrication of two- and three-tier integrated circuits, leading early development of 3D integration.

The chip's heaters emulate both the background levels of heat within a stack and localized hot spots. Hot spots often occur in the most buried and inaccessible areas of a chip stack, making it difficult for 3D-chip developers to assess whether cooling schemes, such as microchannels delivering cold liquid, are reaching those spots and are effective enough.

That's where temperature-sensing elements come in. The chip is distributed with what Chen likens to "tiny thermometers" that read out the temperature in multiple locations across the chip as coolants are applied.

These thermometers are actually diodes, or switches that allow current to flow through a circuit as voltage is applied. As the diodes heat up, the current-to-voltage ratio changes. "We're able to check a diode's performance and know that it's 200 degrees C, or 100 degrees C, or 50 degrees C, for example," Keech says. "We thought creatively about how devices could fail from overheating, and then used those same properties to design useful measurement tools."

Chen and Keech — along with other design, fabrication, and electrical test experts across the laboratory — are now collaborating with HRL Laboratories researchers as they couple the chip with novel cooling technologies, and integrate those technologies into a 3DHI stack that could boost RF signal power. "We need to cool the heat equivalent of more than 190 laptop CPUs [central processing units], but in the size of a single CPU package," Christopher Roper, co-principal investigator at HRL, said in a recent press release announcing their program.

According to Keech, the rapid timeline for delivering the chip was a challenge overcome by teamwork through all phases of the chip's design, fabrication, test, and 3D heterogenous integration.

"Stacked architectures are considered the next frontier for microelectronics," he says. "We want to help the U.S. government get ahead in finding ways to integrate them effectively and enable the highest performance possible for these chips."

The laboratory team presented this work at the annual Government Microcircuit Applications and Critical Technology Conference (GOMACTech), held March 17-20.

To understand what drives disease progression in tissues, scientists need more than just a snapshot of cells in isolation — they need to see where the cells are, how they interact, and how that spatial organization shifts across disease states. A new computational method called MESA (Multiomics and Ecological Spatial Analysis), detailed in a study published in Nature Genetics, is helping researchers study diseased tissues in more meaningful ways.

The work details the results of a collaboration between researchers from MIT, Stanford University, Weill Cornell Medicine, the Ragon Institute of MGH, MIT, and Harvard, and the Broad Institute of MIT and Harvard, and was led by the Stanford team.

MESA brings an ecology-inspired lens to tissue analysis. It offers a pipeline to interpret spatial omics data — the product of cutting-edge technology that captures molecular information along with the location of cells in tissue samples. These data provide a high-resolution map of tissue “neighborhoods,” and MESA helps make sense of the structure of that map.

“By integrating approaches from traditionally distinct disciplines, MESA enables researchers to better appreciate how tissues are locally organized and how that organization changes in different disease contexts, powering new diagnostics and the identification of new targets for preventions and cures,” says Alex K. Shalek, the director of the Institute for Medical Engineering and Science (IMES), the J. W. Kieckhefer Professor in IMES and the Department of Chemistry, and an extramural member of the Koch Institute for Integrative Cancer Research at MIT, as well as an institute member of the Broad Institute and a member of the Ragon Institute.

“In ecology, people study biodiversity across regions — how animal species are distributed and interact,” explains Bokai Zhu, MIT postdoc and author on the study. “We realized we could apply those same ideas to cells in tissues. Instead of rabbits and snakes, we analyze T cells and B cells.”

By treating cell types like ecological species, MESA quantifies “biodiversity” within tissues and tracks how that diversity changes in disease. For example, in liver cancer samples, the method revealed zones where tumor cells consistently co-occurred with macrophages, suggesting these regions may drive unique disease outcomes.

“Our method reads tissues like ecosystems, uncovering cellular ‘hotspots’ that mark early signs of disease or treatment response,” Zhu adds. “This opens new possibilities for precision diagnostics and therapy design.”

MESA also offers another major advantage: It can computationally enrich tissue data without the need for more experiments. Using publicly available single-cell datasets, the tool transfers additional information — such as gene expression profiles — onto existing tissue samples. This approach deepens understanding of how spatial domains function, especially when comparing healthy and diseased tissue.

In tests across multiple datasets and tissue types, MESA uncovered spatial structures and key cell populations that were previously overlooked. It integrates different types of omics data, such as transcriptomics and proteomics, and builds a multilayered view of tissue architecture.

Currently available as a Python package, MESA is designed for academic and translational research. Although spatial omics is still too resource-intensive for routine in-hospital clinical use, the technology is gaining traction among pharmaceutical companies, particularly for drug trials where understanding tissue responses is critical.

“This is just the beginning,” says Zhu. “MESA opens the door to using ecological theory to unravel the spatial complexity of disease — and ultimately, to better predict and treat it.”

The company doesn’t keep logs, so couldn’t turn over data:

Windscribe, a globally used privacy-first VPN service, announced today that its founder, Yegor Sak, has been fully acquitted by a court in Athens, Greece, following a two-year legal battle in which Sak was personally charged in connection with an alleged internet offence by an unknown user of the service.

The case centred around a Windscribe-owned server in Finland that was allegedly used to breach a system in Greece. Greek authorities, in cooperation with INTERPOL, traced the IP address to Windscribe’s infrastructure and, unlike standard international procedures, proceeded to initiate criminal proceedings against Sak himself, rather than pursuing information through standard corporate channels...

Political Retribution for Telling the Truth Weakens the Entire Infosec Community and Threatens Our Democracy; Letter Remains Open for Further Sign-Ons

SAN FRANCISCO – The Trump Administration must cease its politically motivated investigation of former U.S. Cybersecurity and Infrastructure Security Agency Director Christopher Krebs, the Electronic Frontier Foundation (EFF) and dozens of prominent cybersecurity and election security experts urged in an open letter. 

The letter – signed by preeminent names from academia, civil society, and the private sector – notes that security researchers play a vital role in protecting our democracy, securing our elections, and building, testing, and safeguarding government infrastructure. 

“By placing Krebs and SentinelOne in the crosshairs, the President is signaling that cybersecurity professionals whose findings do not align with his narrative risk having their businesses and livelihoods subjected to spurious and retaliatory targeting, the same bullying tactic he has recently used against law firms,” EFF’s letter said. “As members of the cybersecurity profession and information security community, we counter with a strong stand in defense of our professional obligation to report truthful findings, even – and especially – when they do not fit the playbook of the powerful. And we stand with Chris Krebs for doing just that.” 

President Trump appointed Krebs as Director of the Cybersecurity and Infrastructure Security Agency in the U.S. Department of Homeland Security in November 2018, and then fired him in November 2020 after Krebs publicly contradicted Trump's false claims of widespread fraud in the 2020 presidential election. 

Trump issued a presidential memorandum on April 9 directing Attorney General Pam Bondi and Homeland Security Secretary Kristi Noem to investigate Krebs, and directing Bondi and Director of National Intelligence Tulsi Gabbard to revoke security clearances held by Krebs and the cybersecurity company for which he worked, SentinelOne.  EFF’s letter urges that both of these actions be reversed immediately. 

“An independent infosec community is fundamental to protecting our democracy, and to the profession itself,” EFF’s letter said. “It is only by allowing us to do our jobs and report truthfully on systems in an impartial and factual way without fear of political retribution that we can hope to secure those systems. We take this responsibility upon ourselves with the collective knowledge that if any one of us is targeted for our work hardening these systems, then we all can be. We must not let that happen. And united, we will not let that happen.” 

EFF also has filed friend-of-the-court briefs supporting four law firms targeted for retribution in Trump’s unconstitutional executive orders. 

For the letter in support of Krebs: https://www.eff.org/document/chris-krebs-support-letter-april-28-2025

To sign onto the letter: https://eff.org/r.uq1r 

Contact:  WilliamBudingtonSenior Staff Technologistbill@eff.org

Once again, the Texas legislature is coming after the most common method of safe and effective abortion today—medication abortion.

Senate Bill (S.B.) 2880* seeks to prevent the sale and distribution of abortion pills—but it doesn’t stop there. By restricting access to certain information online, the bill tries to keep people from learning about abortion drugs, or even knowing that they exist.

If passed, S.B. 2880 would make it illegal to “provide information” on how to obtain an abortion-inducing drug. If you exchange e-mails or have an online chat about seeking an abortion, you could violate the bill. If you create a website that shares information about legal abortion services in other states, you could violate the bill. Even your social media posts could put you at risk.

On top of going after online speakers who create and post content themselves, the bill also targets social media platforms, websites, email services, messaging apps, and any other “interactive computer service” simply for hosting or making that content available.

In other words, Texas legislators not only want to make sure no one can start a discussion on these topics, they also want to make sure no one can find one. The goal is to wipe this information from the internet altogether. That creates glaring free-speech issues with this bill and, if passed, the consequences would be dire.

The bill is carefully designed to scare people into silence.

First, S.B. 2880 empowers average citizens to sue anyone that violates the law. An “interactive computer service” can also be sued if it “allows residents of [Texas] to access information or material that aids, abets, assists or facilitates efforts to obtain elective abortions or abortion-inducing drugs.”

So, similar to Texas Senate Bill 8, the bill encourages anyone to file lawsuits against those who merely speak about or provide access to certain information. This is intended to, and will, chill free speech. The looming threat of litigation can be used to silence those who seek to give women truthful information about their reproductive options—potentially putting their health or lives in danger.

Second, S.B. 2880 encourages online intermediaries to take down abortion-related content. For example, if sued under the law, a defendant platform can escape liability by showing that, once discovered, they promptly “block[ed] access to any information . . . that assists or facilitates efforts to obtain elective abortions or abortion-inducing drugs.”

The bill also grants them “absolute and nonwaivable immunity” against claims arising from takedowns, denials of service, or any other “action taken to restrict access to or availability of [this] information.” In other words, if someone sues a social media platform or internet service provider for censorship, they are well-shielded from facing consequences. This further tips the scales in favor of blocking more websites, posts, and users.

In three different provisions of the 43-page bill, the drafters go out of their way to assure us that S.B. 2880 should not be construed to prohibit speech or conduct that’s protected by the First Amendment. But simply stating that the law does not restrict free speech does not make it so. The obvious goal of this bill is to restrict access to information about abortion medications online. It’s hard to imagine what claims could be brought under such a bill that don’t implicate our free speech rights.

The bill’s imposition of civil and criminal liability also conflicts with a federal law that protects online intermediaries’ ability to host user-generated speech, 47 U.S.C. § 230 (“Section 230”), including speech about abortion medication. Although the bill explicitly states that it does not conflict with Section 230, that assurance remains meaningful only so long as Section 230’s protections remain robust. But Congress is currently considering revisions—or even a full repeal of Section 230. Any weakening of Section 230 will create more space for those empowered by this bill to use the courts to pressure intermediaries/platforms to remove information about abortion medication.

Whenever the government tries to restrict our ability to access information, our First Amendment rights are threatened. This is exactly what Texas lawmakers are trying to do with S.B. 2880. Anyone who cares about free speech—regardless of how they feel about reproductive care—should urge lawmakers to oppose this bill and others like it.

*H.B. 5510 is the identical House version of S.B. 2880.

Many diseases are caused by a missing or defective copy of a single gene. For decades, scientists have been working on gene therapy treatments that could cure such diseases by delivering a new copy of the missing genes to the affected cells.

Despite those efforts, very few gene therapy treatments have been approved by the FDA. One of the challenges to developing these treatments has been achieving control over how much the new gene is expressed in cells — too little and it won’t succeed, too much and it could cause serious side effects.

To help achieve more precise control of gene therapy, MIT engineers have tuned and applied a control circuit that can keep expression levels within a target range. In human cells, they showed that they could use this method to deliver genes that could help treat diseases including fragile X syndrome, a disorder that leads to intellectual disability and other developmental problems.

“In theory, gene supplementation can solve monogenic disorders that are very diverse but have a relatively straightforward gene therapy fix if you could control the therapy well enough,” says Katie Galloway, the W. M. Keck Career Development Professor in Biomedical Engineering and Chemical Engineering and the senior author of the new study.

MIT graduate student Kasey Love is the lead author of the paper, which appears today in Cell Systems. Other authors of the paper include MIT graduate students Christopher Johnstone, Emma Peterman, and Stephanie Gaglione, and Michael Birnbaum, an associate professor of biological engineering at MIT.

Delivering genes

While gene therapy holds promise for treating a variety of diseases, including hemophilia and sickle cell anemia, only a handful of treatments have been approved so far, for an inherited retinal disease and certain blood cancers.

Most gene therapy approaches use a virus to deliver a new copy of a gene, which is then integrated into the DNA of host cells. Some cells may take up many copies of the gene, while others don’t receive any.

“Simple overexpression of that payload can result in a really wide range of expression levels in the target genes as they take up different numbers of copies of those genes or just have different expression levels,” Love says. “If it's not expressing enough, that defeats the purpose of the therapy. But on the other hand, expressing at too high levels is also a problem, as that payload can be toxic.”

To try to overcome this, scientists have experimented with different types of control circuits that constrain expression of the therapeutic gene. In this study, the MIT team decided to use a type of circuit called an incoherent feedforward loop (IFFL).

In an IFFL circuit, activation of the target gene simultaneously activates production of a molecule that suppresses gene expression. One type of molecule that can be used to achieve that suppression is microRNA — a short RNA sequence that binds to messenger RNA, preventing it from being translated into protein.

In this study, the MIT team designed an IFFL circuit, called “ComMAND” (Compact microRNA-mediated attenuator of noise and dosage), so that a microRNA strand that represses mRNA translation is encoded within the therapeutic gene. The microRNA is located within a short segment called an intron, which gets spliced out of the gene when it is transcribed into mRNA. This means that whenever the gene is turned on, both the mRNA and the microRNA that represses it are produced in roughly equal amounts.

This approach allows the researchers to control the entire ComMAND circuit with just one promoter — the DNA site where gene transcription is turned on. By swapping in promoters of different strengths, the researchers can tailor how much of the therapeutic gene will be produced.

In addition to offering tighter control, the circuit’s compact design allows it to be carried on a single delivery vehicle, such as a lentivirus or adeno-associated virus, which could improve the manufacturability of these therapies. Both of those viruses are frequently used to deliver therapeutic cargoes.

“Other people have developed microRNA based incoherent feed forward loops, but what Kasey has done is put it all on a single transcript, and she showed that this gives the best possible control when you have variable delivery to cells,” Galloway says.

Precise control

To demonstrate this system, the researchers designed ComMAND circuits that could deliver the gene FXN, which is mutated in Friedreich’s ataxia — a disorder that affects the heart and nervous system. They also delivered the gene Fmr1, whose dysfunction causes fragile X syndrome. In tests in human cells, they showed that they could tune gene expression levels to about eight times the levels normally seen in healthy cells.

Without ComMAND, gene expression was more than 50 times the normal level, which could pose safety risks. Further tests in animal models would be needed to determine the optimal levels, the researchers say.

The researchers also performed tests in rat neurons, mouse fibroblasts, and human T-cells. For those cells, they delivered a gene that encodes a fluorescent protein, so they could easily measure the gene expression levels. In those cells, too, the researchers found that they could control gene expression levels more precisely than without the circuit.

The researchers now plan to study whether they could use this approach to deliver genes at a level that would restore normal function and reverse signs of disease, either in cultured cells or animal models.

“There's probably some tuning that would need to be done to the expression levels, but we understand some of those design principles, so if we needed to tune the levels up or down, I think we'd know potentially how to go about that,” Love says.

Other diseases that this approach could be applied to include Rett syndrome, muscular dystrophy and spinal muscular atrophy, the researchers say.

“The challenge with a lot of those is they're also rare diseases, so you don't have large patient populations,” Galloway says. “We're trying to build out these tools that are robust so people can figure out how to do the tuning, because the patient populations are so small and there isn't a lot of funding for solving some of these disorders.”

The research was funded by the National Institute of General Medical Sciences, the National Science Foundation, the Institute for Collaborative Biotechnologies, and the Air Force Research Laboratory. 

The EPA administrator's strategy for dismantling the scientific justification for climate rules could focus on the costs to industry.

The International Code Council is expected to vote Monday on a proposal for reducing flood damage in new construction.

Legislators are exploring ways to bring air conditioning to the state's prison system, where only about a third of detainees have regular access to cool air.

The industry has lost a record $6.9 billion in investment this year as companies navigate an "unstable policy environment," according to a new analysis.

The influential lobbying group for CEOs argued that the shareholder resolution process has been "co-opted" by environmental activists.

President Donald Trump’s tariff war has left investors struggling to navigate a fire hose of headlines from the White House, most of which have added to the uncertainty gripping markets.

Elliott Investment Management has made clear it wants BP to return to its core oil and gas business.

A “certain major country” is keen on unilateralism and protectionism and has caused “serious impacts” on international rules and order, Chinese President Xi Jinping said.

Companies like BYD, Great Wall, Geely and Chery Automobile are reaching outward as they build the scale they need to survive cut-throat competition in their home market.

The federal government is using social media surveillance to target student visa holders living in the United States for online speech the Trump administration disfavors. This new program, called “Catch and Revoke,” is an effort to revoke visas and appears to be a cross-agency collaboration between the State Department, the Department of Homeland Security (DHS), and the Department of Justice. It includes a dedicated task force and the use of AI and other data analytic tools to review the public social media accounts of tens of thousands of student visa holders. Though the full scope remains unclear, current reports indicate that the administration is surveilling for “pro-Hamas” sentiment, “antisemitic activity,” or even just “conduct that bears a hostile attitude toward U.S. citizens or U.S. culture.” At the time of publishing of this blog post, the federal government has already revoked over 1600 student visas for a variety of reasons.

This social media surveillance program is an alarming attack on freedom of speech and privacy—for both visa holders here in the United States and their American associates.

A Dangerous Erosion of Free Speech

While there is some nuance in the interplay between freedom of speech and immigration law, one principle remains evident: foreign nationals who currently reside in the U.S.—including student visa holders—are protected by the First Amendment. The Supreme Court stated in Bridges v. Wixon (1945) that “[f]reedom of speech and of press is accorded aliens residing in this country.”

First Amendment-Protected Political Speech

The Trump administration’s efforts to revoke student visas based, in part, on what students have said publicly on social media is especially constitutionally problematic given that the government is targeting core First Amendment-protected political speech. As the Supreme Court stated in Mills v. Alabama (1966), a central purpose of the First Amendment is to “protect the free discussion of governmental affairs,” whether on political issues, public officials, or how the government should operate.

The administration is targeting non-citizen students for “pro-Hamas,” antisemitic, and even just pro-Palestinian speech. Yet what falls under these categories is vague and not clearly defined. For example, the administration detained a Georgetown University researcher due to social media posts that are critical of Israel, but do not express support for Hamas.

More importantly, even controversial or offensive speech falls within the protections of the First Amendment. There are several categories of speech that do not enjoy First Amendment protection, including true threats of violence, inciting imminent violence, and providing material support for terrorism. However, short of rising to that level, the student speech targeted by the administration is protected by the First Amendment. Worse still, the administration is broadly going after students who simply appear to be “social activists” or are engaged in speech that is generically “anti-American.”

Such an overbroad social media surveillance and visa revocation program—one that sweeps in wholly lawful speech—strikes at the heart of what the First Amendment was intended to protect against.

Chilling Effect

Social media surveillance motivated by the government’s desire to punish political speech will chill (and surely has already chilled) student visa holders from speaking out online.

The Supreme Court stated in Lamont v. Postmaster General (1965) that a government policy that causes individuals “to feel some inhibition” in freely expressing themselves “is at war with the ‘uninhibited, robust, and wide-open’ debate and discussion that are contemplated by the First Amendment.” More recently, Supreme Court Justice Sotomayor expressed in a concurring opinion that “[a]wareness that the Government may be watching chills associational and expressive freedoms” guaranteed by the First Amendment.

That is, student visa holders are more likely to engage in self-censorship and refrain from expressing dissenting or controversial political views. Or they may choose to disengage from social media entirely, to avoid the risk that even seemingly benign posts will affect their visa status and their ability to continue their education in the United States.

Student visa holders may also limit whom they connect with on social media, particularly if they fear those connections will have political views the current administration doesn’t like. The administration has not expressly stated that it will limit its surveillance only to the social media posts of student visa holders, which means it may also look at posts made by those in the students’ networks. This, too, undermines the First Amendment. The freedom to associate and express political views as a group—“particularly controversial ones”—is a fundamental aspect of freedom of speech, as the Supreme Court stated in its landmark NAACP v. Alabama (1958) decision.

American Citizens Impacted

Because student visa holders’ social networks undoubtedly include U.S. citizens, those citizens may also be subject to social media scrutiny, and therefore will also be chilled from freely speaking or associating online. Government agents have previously held visa holders responsible for the activity of their social media connections. Knowing this, a U.S. citizen who has a non-citizen friend or family member in the U.S. on a student visa might hesitate to post criticisms of the government—even if fully protected by the First Amendment—fearing the posts could negatively impact their loved one. A general climate of government surveillance may also lead U.S. citizens to self-censor on social media, even without any foreign national friends or family.

A Threat to Digital Privacy

Social media surveillance, even of publicly available profiles and especially with automated tools, can invade personal privacy. The Supreme Court has repeatedly held that the government’s collection and aggregation of publicly available personal information—particularly when enhanced by technology—can implicate privacy interests. The government can obtain personal information it otherwise would not have access to or that would usually be difficult to find across disparate locations.

Social media aggregates personal information in one place, including some of the most intimate details of our lives, such as our health information, likes and dislikes, political views and religious beliefs, and people with whom we associate. And automated tools can easily search for and help find this information. Even people who choose not to post much personal information on social media might still be exposed by comments and tags made by other users.

Constitutional Harms are Exacerbated by Automated Tools

The Trump administration is reportedly deploying artificial intelligence and other automated tools to assist in its review of student visa holders’ social media posts. While facts are still coming to light, any form of automation is likely to amplify speech and privacy harms to student visa holders.

By the government’s own assessment in another context—evaluating the admissibility of visa applicants (discussed below)—social media surveillance has not proven effective at assessing security threats.

Human review of public social media posts is itself prone to problems. Social media posts are highly context-specific, and government officials often have trouble differentiating between sarcasm, parody, and exaggeration from unlawful support for controversial causes. This leads to mistakes and misinterpretations. For example, in 2012 an Irish citizen was turned back at the border because DHS agents misinterpreted two of his Twitter posts: one, that he was going to “destroy America” – slang for partying – and two, that he was going to “dig up Marilyn Monroe’s grave” – a joke. These mistakes are even more likely when the posts are not in English or when they contain cultural references .

Human review augmented by automated tools is just as bad. Automated tools also have difficulty understanding the nuances of language, as well as the broader context in which a statement was made. These algorithms are also designed to replicate patterns in existing datasets, but if the data is biased, the technology simply reinforces those biases. As such, automated tools are similarly prone to mistakes and misinterpretations. Yet people often defer to automated outputs thinking they are correct or fair simply because a computer was used to produce them. And in some cases, decision-makers may even use these tools to justify or cover their own biases.

Most concerning would be if automated systems were permitted to make final visa revocation decisions without any human review. As EFF has repeatedly stated, automated tools should never get the final say on whether a person should be policed, arrested, denied freedom, or, in this case, stripped of a student visa and forcibly barred from completing their education.

Government Social Media Surveillance is Not New—and is Expanding

That the Trump administration is using social media surveillance on student visa holders residing in the United States is a disturbing apparent escalation of a longstanding trend.

EFF has long sounded the alarm on the civil liberty harms of government social media surveillance. In particular, since 2019, visa applicants have been required to disclose all social media accounts they have used in the last five years to the U.S. government. That policy is the subject of an ongoing lawsuit, Doc Society v. Pompeo, in which EFF filed an amicus brief.

Secretary of State Marco Rubio recently upped the ante by ordering officials to deny visas to new or returning student applicants if their social media broadly demonstrates “a hostile attitude toward U.S. citizens or U.S. culture (including government, institutions, or founding principles).” Notably, Rubio indicated this standard could also apply to current student visa holders. The State Department also announced it will review the social media of any visa applicant who has been to Gaza since 2007.

The Trump administration has also proposed dramatically expanding social media scrutiny by requiring non-citizens already legally residing in the U.S. to disclose social media accounts on a variety of forms related to immigration benefits, such as people seeking lawful permanent residency or naturalization. U.S. Citizenship and Immigration Services (USCIS), a component of DHS, also announced it would look for “antisemitic activity” on social media to deny immigration benefits to individuals currently in the country.

Protecting Your Accounts

There are general steps you can take to better protect your social media accounts from surveillance. Understand, however, that the landscape is shifting rapidly and not all protections are foolproof. Law enforcement may be able to get a warrant for your private information and messages if a judge is convinced there is preliminary evidence supporting probable cause of criminal activity. And non-governmental individuals and groups have recently used other forms of technology like face recognition to identify and report student activists for potential deportation. You should conduct your own individualized risk assessment to determine what online activity is safe for you.

Still, it never hurts to better secure your online privacy. For your current social media accounts, consider locking them down:

• Make public accounts private and ensure only approved connections can see your content. Note that if your past public posts have already been copied and saved by an outside party, making your account private will not undo this. It will, however, better protect your future posts.
• Some platforms make certain information publicly viewable, even if you’ve made your account private. Other information may be public by default, but can be made private. Review each platform’s privacy settings to limit what information is shared publicly, including friend lists, contact information, and location information.
• You should also review your friends or followers list to ensure you know every person you’ve approved, especially when making a once-public account private.

If you create a new social media account:

• Query whether you want to attach your legal name to it. Many platforms allow you to have a pseudonymous account.
• When setting up the account, don’t provide more personal information than is necessary.

EFF’s Surveillance Self-Defense guide provides additional information on protecting your social media accounts from a variety of actors. If you're not sure what information is publicly available about you on social networks or other sites, consider doing some research to see what, if anything, others would find.

By targeting international students for broad categories of online speech, this administration is fostering a climate of fear, making students anxious that a single post or errant “like” could cost them their U.S. visa or even lead to detention and deportation. This will, ultimately, stifle political debate and silence dissent–for non-citizens and citizens alike–undermining the open dialogue crucial to democracy.

Researchers from the Critical Analytics for Manufacturing Personalized-Medicine (CAMP) interdisciplinary research group of the Singapore-MIT Alliance for Research and Technology (SMART), MIT’s research enterprise in Singapore, in collaboration with MIT, A*STAR Skin Research Labs, and the National University of Singapore, have developed a novel method that can quickly and automatically detect and monitor microbial contamination in cell therapy products (CTPs) early on during the manufacturing process. By measuring ultraviolet light absorbance of cell culture fluids and using machine learning to recognize light absorption patterns associated with microbial contamination, this preliminary testing method aims to reduce the overall time taken for sterility testing and, subsequently, the time patients need to wait for CTP doses. This is especially crucial where timely administration of treatments can be life-saving for terminally ill patients. Cell therapy represents a promising new frontier in medicine, especially in treating diseases such as cancers, inflammatory diseases, and chronic degenerative disorders by manipulating or replacing cells to restore function or fight disease. However, a major challenge in CTP manufacturing is quickly and effectively ensuring that cells are free from contamination before being administered to patients. Existing sterility testing methods, based on microbiological methods,  are labor-intensive and require up to 14 days to detect contamination, which could adversely affect critically ill patients who need immediate treatment. While advanced techniques such as rapid microbiological methods (RMMs) can reduce the testing period to seven days, they still require complex processes such as cell extraction and growth enrichment mediums, and they are highly dependent on skilled workers for procedures such as sample extraction, measurement, and analysis. This creates an urgent need for new methods that offer quicker outcomes without compromising the quality of CTPs, meet the patient-use timeline, and use a simple workflow that does not require additional preparation. In a paper titled “Machine learning aided UV absorbance spectroscopy for microbial contamination in cell therapy products,” published in the journal Scientific Reports, SMART CAMP researchers described how they combined UV absorbance spectroscopy to develop a machine learning-aided method for label-free, noninvasive, and real-time detection of cell contamination during the early stages of manufacturing. This method offers significant advantages over both traditional sterility tests and RMMs, as it eliminates the need for staining of cells to identify labelled organisms, avoids the invasive process of cell extraction, and delivers results in under half-an-hour. It provides an intuitive, rapid “yes/no” contamination assessment, facilitating automation of cell culture sampling with a simple workflow. Furthermore, the developed method does not require specialized equipment, resulting in lower costs. “This rapid, label-free method is designed to be a preliminary step in the CTP manufacturing process as a form of continuous safety testing, which allows users to detect contamination early and implement timely corrective actions, including the use of RMMs only when possible contamination is detected. This approach saves costs, optimizes resource allocation, and ultimately accelerates the overall manufacturing timeline,” says Shruthi Pandi Chelvam, senior research engineer at SMART CAMP and first author of the paper. “Traditionally, cell therapy manufacturing is labor-intensive and subject to operator variability. By introducing automation and machine learning, we hope to streamline cell therapy manufacturing and reduce the risk of contamination. Specifically, our method supports automated cell culture sampling at designated intervals to check for contamination, which reduces manual tasks such as sample extraction, measurement, and analysis. This enables cell cultures to be monitored continuously and contamination to be detected at early stages,” says Rajeev Ram, the Clarence J. LeBel Professor in Electrical Engineering and Computer Science at MIT, a principal investigator at SMART CAMP, and the corresponding author of the paper. Moving forward, future research will focus on broadening the application of the method to encompass a wider range of microbial contaminants, specifically those representative of current good manufacturing practices environments and previously identified CTP contaminants. Additionally, the model’s robustness can be tested across more cell types apart from MSCs. Beyond cell therapy manufacturing, this method can also be applied to the food and beverage industry as part of microbial quality control testing to ensure food products meet safety standards.