Background: Large language models use machine learning to produce natural language. These models have a range of potential applications in health care, such as patient education and diagnosis. However, evaluations of large language models in health care are still scarce. Objective: This study aimed to (1) evaluate the accuracy and efficiency of automated fact-checking by 2 large language models and (2) illustrate a process through which a large language model might support a patient in redrafting a prompt to include key information needed for patient safety. Methods: A parallel comparison of 2 large language models and 3 human experts was conducted. A clinical scenario was devised in which a woman aged 23 years questions the safety of retinoid treatment for acne by sending prompts to 2 large language models (GPT-4o and OpenBioLLM-70B). GPT-4o and OpenBioLLM-70B were asked to suggest improvements to the patient’s initial prompt to elicit key information for clinical decision-making. After the patient sent the revised prompt to the large language models, the models were then asked to fact-check the final response. To test the generalizability of automated fact-checking, a set of 20 clinical statements on disparate topics, mostly related to drug indications, contraindications, and side effects, was developed. The large language models also fact-checked these 20 medical statements. The results were compared against the evaluations of 3 clinical experts. The outcome measures were as follows: (1) percentage of accuracy of automated fact-checking, (2) time to complete fact-checking, and (3) a binary outcome for prompt redrafting (advising the patient to revise her prompt by naming her acne medication to address safety concerns). Results: For the scenario of a patient with acne, GPT-4o and OpenBioLLM-70B both had 86% agreement with the clinical experts’ fact-checking. The large language models did not consistently convey the urgency of discontinuing isotretinoin treatment when pregnancy is suspected. In addition, the models did not adequately convey the importance of folic acid supplementation during pregnancy. For the set of 20 medical claims, GPT-4o fact-checking had 100% agreement with that of human experts, whereas OpenBioLLM-70B had 95% agreement. OpenBioLLM-70B diverged from human experts and GPT-4o on 1 question related to pediatric use of antihistamines. The expert fact-checks took a mean time of 18 (SD 3.74) minutes, GPT-4o took 42 seconds, and OpenBioLLM-70B took 33 minutes. The GPT-4o responses for the acne scenario had some inconsistencies but zero fabrication and no obvious omissions. In contrast, OpenBioLLM-70B omitted 1 key information item needed for patient safety. Conclusions: GPT-4o can interact with patients to improve the quality and comprehensiveness of the information contained in health-related prompts. GPT-4o and OpenBioLLM-70B can conduct efficient fact-checking that is close to the level of accuracy of human experts. Human experts need to perform additional checks for accuracy and safety.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/35585a0a05dacf1d0cdf467d36579925" />
Elraglusib Delivers Rare Survival Benefit in Pancreatic Cancer
A randomized Phase II trial offers a rare signal of progress in pancreatic cancer, a disease long marked by therapeutic stagnation, with investigators reporting that the experimental agent elraglusib (9-ING-41) improved survival when added to standard chemotherapy of gemcitabine plus nab-paclitaxel (GnP).
Published in Nature Medicine, the study evaluated the novel drug—developed within an academic setting at Northwestern University—in patients with metastatic pancreatic cancer. The findings suggest that targeting glycogen synthase kinase-3 beta (GSK-3β), a protein not previously exploited clinically in this disease, could open a new therapeutic avenue.
“This is one of the first trials in a randomized setting that has been positive in pancreatic cancer in the last decade,” said Devalingam Mahalingam, MD, PhD, a study leader at Northwestern University. “There was really a barren spell… many failed trials. So it’s nice to see a positive trial.”
A modest but meaningful survival gain
The multicenter trial enrolled 233 patients across North America and Europe, randomly assigning them to receive chemotherapy alone or in combination with elraglusib. Patients receiving the combination lived a median of 10.1 months compared with 7.2 months for chemotherapy alone, and the addition of elraglusib reduced the risk of death by 38%.
Perhaps more striking, survival at one year doubled in the experimental arm (44% vs. 22%), and approximately 13% of patients remained alive at two years—an uncommon outcome in metastatic pancreatic cancer.
Mahalingam emphasized that the benefit was not necessarily reflected in higher tumor response rates. Instead, patients appeared to derive prolonged disease control.
“We didn’t really see much more tumor shrinkage compared to chemo alone,” he said. “But patients stayed on the treatment arm longer… sometimes they would reduce or drop the chemo and just stay on the drug.”
This pattern, he added, points toward a mechanism beyond direct cytotoxicity.
A different mechanism of action
Unlike conventional chemotherapy, which primarily targets rapidly dividing cells, elraglusib appears to act on the tumor microenvironment—the complex ecosystem of immune cells, stromal tissue, and signaling molecules surrounding cancer cells.
The drug inhibits GSK-3β, a protein involved in multiple cellular processes, including metabolism, signaling pathways, and immune regulation. While broadly expressed in normal tissues, GSK-3β can be co-opted by tumors to promote growth and suppress immune responses.
“GSK-3 beta is expressed in many tumors,” Mahalingam said. “It’s part of a central regulator of normal cell functioning… but in cancer, this pathway is used to allow for tumor growth and proliferation.”
Preliminary analyses from the trial suggest that elraglusib may enhance antitumor immunity. Biopsies and blood-based markers indicated changes consistent with immune activation, supporting the hypothesis that the drug helps re-engage the immune system in a tumor type typically resistant to immunotherapy.
“We saw what we call immunomodulatory effects,” Mahalingam noted. “The immune cells might be driving some of the survival benefit we see.”
This is particularly notable given the long-standing failure of checkpoint inhibitors in pancreatic cancer, a tumor characterized by a highly immunosuppressive microenvironment.
Developed in academia
Elraglusib’s development trajectory also sets it apart. The compound originated in academic laboratories more than a decade ago, with early work spanning Northwestern University, the University of Illinois Chicago, and the Mayo Clinic.
“This is what not many drugs are—developed within an academic setting,” Mahalingam said. “It was founded in a chemistry lab… and then moved into a spin-off company to raise funding for trials.”
After preclinical development between 2010 and 2015, the drug entered early-phase trials around 2017 and has since been evaluated across multiple tumor types, with a recent focus on pancreatic cancer.
The randomized Phase II trial marks the first time a GSK-3β inhibitor has demonstrated efficacy beyond early-stage testing.
Broad eligibility, real-world relevance
Investigators designed the study with relatively broad inclusion criteria, enrolling patients with high tumor burden and poor nutritional status—characteristics common in real-world pancreatic cancer populations but often excluded from clinical trials.
“We allowed patients with very large volumes of disease,” Mahalingam said. “We did not restrict patients for albumin… we didn’t engineer the study to look better.”
This approach may partly explain the modest median survival difference, as some patients progressed too quickly to benefit. However, it also strengthens the generalizability of the findings.
Safety and next steps
Side effects associated with elraglusib were generally manageable and consistent with chemotherapy, including hematologic toxicities, fatigue, and reversible vision changes.
The next step will be a confirmatory Phase III trial, with discussions ongoing with regulators.
“We really need to confirm the studies in a large phase three trial,” Mahalingam said, adding that trial design considerations include how to integrate emerging therapies such as KRAS inhibitors.
Investigators are also exploring combination strategies, including pairing elraglusib with immunotherapy or alternative chemotherapy regimens. Early safety studies suggest such combinations are feasible, though efficacy data remain limited.
Potential beyond pancreatic cancer
Given the central role of GSK-3β in multiple cellular pathways, researchers are already investigating the drug’s potential in other malignancies, including hematologic cancers and pediatric tumors such as Ewing sarcoma and medulloblastoma.
“This is a new class of potential cancer therapeutics,” Mahalingam said. “Certainly, there would be excitement in seeing where this target can be applied to other tumors.”
While challenges remain, the trial’s results offer cautious optimism in a field where progress has been incremental at best.
“Even if it means that this class of drugs can be used for future drug development,” Mahalingam said, “it gives an opportunity to expand therapeutic potential—not just for pancreatic cancer, but beyond.”
The post Elraglusib Delivers Rare Survival Benefit in Pancreatic Cancer appeared first on Inside Precision Medicine.
Public Perceptions of AI in Medicine and Implications for Future Medical Education: Cross-Sectional Survey
Background: The integration of artificial intelligence (AI) into clinical practice is contingent on public trust. This trust often depends on physician oversight, yet a significant gap exists between the need for AI-competent physicians and the current state of medical education. While the perspectives of students and experts on this gap are known, the views of the US general public remain largely unquantified. Objective: This study aimed to assess US public perceptions regarding AI in medicine and the corresponding emergent needs for medical education. We specifically sought to quantify public trust in different diagnostic scenarios, concerns about physician overreliance on AI, support for mandatory AI education, and priorities for the future focus of medical training. Methods: We conducted a cross-sectional, web-based survey of adults in the United States in November 2025. Participants (N=524) were recruited via SurveyMonkey Audience. We calculated descriptive statistics, frequencies, proportions (percentages), and 95% CIs for all main survey items. Results: A total of 524 participants completed the survey. Most (n=329, 62.8%; 95% CI 58.6%‐66.9%) placed the most trust in a physician’s diagnosis based on their expertise alone; only 7.8% (n=41; 95% CI 5.5%‐10.1%) trusted an AI-first diagnostic model. Trust was highly contingent on training: 93.9% (n=492) of participants rated formal physician training on AI limitations as “essential” or “very important.” Widespread concern about physician overreliance on AI was reported, with 81.1% (n=425) being “very concerned” or “extremely concerned.” Consequently, 85.1% (n=446) agreed or strongly agreed that training on AI use, ethics, and limitations should be mandatory in medical school. When asked about future educational priorities, 70.2% (n=368; 95% CI 66.3%‐74.1%) believed that medical education should focus on human-centered skills (eg, empathy and communication) over clinical skills. Conclusions: The US public expressed conditional trust in medical AI, strongly preferring physician-led and critically supervised models. These findings reveal a clear public mandate for medical education reform. The public expects future physicians to be mandatorily trained to appraise AI, understand its limitations, and refocus their professional development on the human-centered skills that technology cannot replace.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/3874abd2d5f25c78f21987a16f3af6be" />
ThermoCas9: Gene Editor Targets Cells with Disease-Related Hypomethylation
Research led by Wageningen University in the Netherlands and the Van Andel Institute (VAI) in Michigan has shown that ThermoCas9, a variant of CRISPR, can distinguish tumor DNA from healthy DNA and selectively cut only the former, marking a potential step toward a highly precise cancer therapy.
The method relies on DNA methylation, a process in which methyl groups are added to DNA to regulate whether genes are on or off. In cancer cells, DNA methylation is altered and can therefore act as a molecular “fingerprint” that differentiates tumor cells from healthy ones.
“ThermoCas9 is the first CRISPR-associated enzyme to respond to differences in the most abundant type of DNA methylation in human and other eukaryotic cells,” explained co-senior author John van der Oost, PhD, from Wageningen University. “This means we now have a system that we can target specifically toward tumor cells.”
The study, published in Nature, represents the first time a CRISPR-based method has relied on methylation to target human cancer cells.
“ThermoCas9 uses methylation like an address to precisely target cancer cells while leaving healthy cells untouched,” added co-senior author Hong Li, PhD, from VAI. “The findings could be a game changer.”
After analyzing ThermoCas9’s structure and finding that it can distinguish between unmethylated and methylated genes, Li and team introduced the enzyme into different types of healthy human cells with distinct methylation landscapes and into breast and colorectal cancer cells.
They found that ThermoCas9 cut DNA in the tumor cells while leaving healthy DNA intact, suggesting that the system can detect subtle chemical differences between healthy and tumor cells and act on them.
“ThermoCas9 is a perfect example of the value of fundamental research; you have to know how these individual pieces work together,” said Li. “We used biochemistry and structural biology to discover a mechanism that we one day hope will lead to more precise, effective cancer treatment.”
Although the study highlights the potential of ThermoCas9 as a cancer treatment, it does not show that the selective DNA damage it inflicts leads to tumor cell death. The researchers next steps will focus on damaging tumor DNA sufficiently to trigger cell death.
Of note, aberrant methylation patterns also play a role in diseases other than cancer, including autoimmune disorders. It is therefore possible that ThermoCas9 or a similar CRISPR tool could evolve into a versatile molecular strategy that recognizes diseased cells by their chemical “signature” and selectively disables them.
The post ThermoCas9: Gene Editor Targets Cells with Disease-Related Hypomethylation appeared first on Inside Precision Medicine.
Ultra- and Diafiltration Clear Leachables Effectively
In the push to de-risk biologics manufacturing, downstream purification steps are increasingly under the microscope. Now, new research led by Jonathan Bones, PhD, principal investigator in the characterization and comparability group at the National Institute for Bioprocessing Research in Dublin, and his colleagues provided compelling evidence that ultrafiltration and diafiltration (UF/DF) deliver robust clearance of process-related leachables—while also offering a predictive framework to better understand that performance.
Although UF/DF has long been assumed to reduce small-molecule contaminants, systematic data have been scarce. To address this gap, the team evaluated 28 representative organic compounds spiked into three distinct protein systems. Using liquid chromatography–high resolution mass spectrometry, they tracked how effectively these compounds were removed during UF/DF operations.
The results were striking. Twenty-four of the compounds demonstrated greater than 98% clearance across all three protein processes. Notably, variations in protein characteristics and process parameters had minimal impact on removal efficiency. Instead, clearance behavior was remarkably consistent, as reflected in similar sieving coefficients across the systems.
The intrinsic physicochemical properties of the leachables impacted clearance. Among these, lipophilicity—expressed as the octanol-water partition coefficient (Log P)—emerged as the dominant factor. Compounds with Log P values below four exhibited near-ideal clearance, while even highly hydrophobic molecules (Log P above seven) still achieved removal rates exceeding 93%. Molecular weight, polarizability, and solvent-accessible surface area also contributed to clearance outcomes.
Beyond empirical findings, the study advances the field with predictive modeling. By applying orthogonal partial least squares (OPLS) regression, the researchers developed tools capable of estimating sieving coefficients based on compound properties. These models could prove invaluable for anticipating leachable behavior without exhaustive experimental testing.
The implications are significant. As regulatory scrutiny around extractables and leachables intensifies, demonstrating effective clearance becomes central to product safety. This work not only confirms that UF/DF is a powerful mitigation step but also equips developers with quantitative tools to support risk assessments.
In an industry where unseen contaminants can pose outsized risks, the ability to both measure and predict their removal marks a meaningful step forward.
The post Ultra- and Diafiltration Clear Leachables Effectively appeared first on GEN – Genetic Engineering and Biotechnology News.
Earlier Surveillance Warranted for Young Cancer Survivors
Approximately one in six adolescents and young adults who survive cancer will be diagnosed with another cancer within 30 years of their initial diagnosis, making their cancer risk more than double that of the general population, shows data from Canada.
“When combined with the relatively high survival rates in this age range, at approximately 86%, there is a growing population of young cancer survivors that will be adversely affected by their cancer diagnosis and its treatment even decades later,” write Miranda Fidler-Benaoudia, a cancer epidemiologist at the University of Calgary Cumming School of Medicine and Cancer Care Alberta, and co-authors in the Canadian Medical Association Journal.
“Given that subsequent primary neoplasms are major contributors to morbidity and premature mortality, these findings underscore the need for innovative solutions to prevent, detect, and treat subsequent primary neoplasms among survivors of adolescent and young adult cancer,” they say.
Although global data suggest that adolescent and young adult cancer survivors are 1.6 to 4.3 times more likely to develop a subsequent cancer than expected in the general population, data are limited for Canada, where cancer rates among this group increased annually by 1.3% from 1998 to 2012, with an estimated 8739 adolescents and young adults diagnosed with cancer in 2022.
The Alberta Adolescent and Young Adult Cancer Survivor Study retrospectively reviewed data for 24,459 people with a neoplasm first diagnosed between 1983 and 2017 at age 15 to 39 years. Of these, 1442 (5.9%) had subsequent primary neoplasms, 1129 (7.6%) of which occurred among the 14,818 people who were five-year survivors.
The researchers report that, overall, adolescent and young adult cancer survivors were 2.2 times more likely to develop a subsequent primary neoplasm than would be expected in the general population, equating to 31.7 excess neoplasms per 10,000 person–years.
For five-year survivors, the incidence was twofold higher versus the general population with an excess risk of 35.7 per 10,000 person–years.
The greatest excess risks occurred among survivors of breast cancer (85.8 per 10,000 person–years), lymphomas other than Hodgkin lymphoma or non-Hodgkin lymphoma (82.0 per 10,000 person–years), and oral cavity, lip, and pharyngeal cancer (74.7 per 10,000 person–years).
Conversely, no significant excesses were observed for survivors of acute myeloid leukemia, central nervous system cancers, ovarian cancer, nonovarian and nontesticular gonadal and related tumors, stomach cancer, lung, bronchial, and tracheal cancer, and endometrial cancer.
After five-year survival, the 30-year cumulative incidence of a subsequent primary neoplasm was 17.7% overall, with incidence highest among survivors of cancers of the oral cavity, lip, or pharynx (28.9%), breast cancer (27.3%), colon cancer (23.5%), and Hodgkin lymphoma (22.7%).
Importantly, the team found that the absolute difference in cumulative incidence between the survivors and general population grew as time since diagnosis increased; for example, for five-year survivors of breast cancer, the absolute difference was 2.9% at 10 years postdiagnosis compared with 12.7% at 30 years postdiagnosis.
“Although people with nearly all types of adolescent and young adult cancer investigated were at an increased risk of developing a subsequent primary neoplasm, survivors of Hodgkin lymphoma and breast cancer were identified as particularly vulnerable populations, with nearly one-third of subsequent primary neoplasms occurring after five-year survival diagnosed in these survivor groups,” Fidler-Benaoudia et al remark. “These findings are consistent with previous studies and reflect the established late effects of radiotherapy, chemotherapy, and hormone therapy.”
However, they also note that genetic factors may play a role, and that genetic counseling and education on maintaining healthy lifestyles are important in caring for survivors.
The most common subsequent primary neoplasms were breast (27.1%), digestive (11.9%), hematopoietic (10.6%), or respiratory (6.8%) cancers, together accounting for nearly 60% of subsequent primary neoplasms.
Fidler-Benaoudia and colleagues say that their findings “offer valuable directions for research and cancer control.”
They point out that at present there are no risk-based survivorship care guidelines that cover the entire age range of the adolescent and young adult cancer population.
“Although both pediatric and adolescent and young adult cancer survivors are at increased risks for late effects, the magnitude of these risks varies and implications for care can differ; thus, clinical investigations that assess the association of treatment exposures with late effects like subsequent primary neoplasms should be a research priority,” the authors write.
They also stress that further work is needed to expand inclusion criteria for early high-risk screening initiatives and to develop specialized treatment guidelines that balance curing the subsequent primary neoplasm while effectively managing the survivors’ late effects.
The post Earlier Surveillance Warranted for Young Cancer Survivors appeared first on Inside Precision Medicine.
Development of a Contextualized, Research-Based Flemish Assessment Framework for Digital Care, Assistance, and Support: Delphi Study
<strong>Background:</strong> The rapid evolution of digital technologies has transformed health, mental health, and social care, offering new modalities of digital care, assistance, and support through web-based platforms, mobile apps, extended reality, wearables, and artificial intelligence systems. Despite this proliferation, there is little consensus on what constitutes “high-quality” digital care. Challenges persist regarding data security, interoperability, accessibility, sustainability, and professional competence, whereas existing standards and regulations provide fragmented guidance. <strong>Objective:</strong> This study aimed to develop a contextualized, consensus-based quality assessment framework for digital care, assistance, and support in Flanders, Belgium. For this purpose, perspectives across technology, organizational processes, and professional competencies were integrated. <strong>Methods:</strong> The study used a multiphase design comprising (1) 10 expert interviews with Flemish government officials; (2) a narrative literature review of 303 peer-reviewed and gray literature sources; (3) a 3-round Delphi study with 50 experts across 5 domains (end users, facilitators, technology developers, deontology and ethics experts, and digital inclusion and media literacy experts); and (4) 4 complementary focus groups and 3 interviews with specialists in artificial intelligence, regulation, social work, mental health, and IT. The Delphi rounds gathered iterative feedback through open-ended elicitation, structured rating, and classification of quality criteria. Quantitative data were analyzed using descriptive statistics, whereas qualitative feedback was subjected to thematic analysis. <strong>Results:</strong> A total of 50 experts participated in round 1, a total of 40 (80%) participated in round 2, and 27 (54%) participated in round 3. Round 1 generated 577 unique quality criteria, consolidated into 26 clusters organized under 3 pillars: technology, organization, and professional competencies. The relative importance across pillars was balanced (mean score 37.29, SD 12.38 for technology; 33.33, SD 10.39 for professional competencies; and 29.80, SD 10.45 for organizations). Accessibility, reliability, and safety ranked highest for the technology; vision, quality monitoring, and infrastructure ranked highest for organization; and support, digital competencies, and ethics ranked highest for professional competencies. The finalized framework included 112 criteria, of which 35 (31.3%) were designated as optional and 77 (68.8%) were designated as minimum requirements. Focus groups and interviews validated the framework’s comprehensiveness and usability, emphasizing proportional implementation, user centrality, and alignment with European Union regulations. Stakeholders highlighted the need for tools, training, and governance mechanisms to ensure adoption and sustainability. <strong>Conclusions:</strong> This study produced a codeveloped, context-sensitive quality assessment framework that balances technological robustness, organizational readiness, and professional competence in digital care, assistance, and support. The framework can serve both as a quality safeguard and a developmental road map. Accompanying self-assessment and governance tools enhance practical applicability. Implementation success will depend on governmental support, resource allocation, and structured feedback loops. Future research should pilot the framework in real-world settings, assess its impact, and establish mechanisms for continuous updates to maintain relevance in a rapidly evolving digital landscape. <strong>Trial Registration:</strong>
<![CDATA[Experts share more on research developments of remternetug for Alzheimer disease.]]>
STAT+: Pharmalittle: We’re reading about FDA seeking more data on a Lilly obesity pill, a pharma 340B win, and more
Top of the morning to you. The middle of the week is upon us and, since you made it this far, why not forge ahead? After all, there is always light at the end of the proverbial tunnel. You never know what you may accomplish. So please join us as we celebrate this notion with a cup or three of delicious stimulation. Our choice today is chocolate raspberry. Meanwhile, we have assembled the latest menu of tidbits to help you along. So please dig in. Have a smashing day, and please feel free to forward any secrets you come across. Our “in basket” is always open. …
The U.S. Food and Drug Administration asked Eli Lilly for more data on liver injury linked to its newly approved obesity pill, Reuters says, citing a letter posted on the agency website. The April 1 letter also said Lilly must conduct post-marketing trials to assess risks related to cardiovascular events and delayed gastric emptying. The drugmaker is required to also conduct a milk-only lactation study in lactating women who have received a dose of the pill to assess concentrations of the drug in breast milk using a validated assay. The weight loss pill, branded Foundayo, a once-daily oral medication that targets the GLP-1 hormone, won approval earlier this month under the Commissioner’s National Priority voucher program, which aims to speed FDA decisions on drugs deemed critical to public health or national security.
AbbVie, Novartis, AstraZeneca, and the Pharmaceutical Research & Manufacturers of America, the industry trade group, notched a victory after a U.S. appeals court vacated an order rejecting their request to block a Maryland drug discount law, remanding the decision for review, Bloomberg Law reports. The U.S. Court of Appeals for the Fourth Circuit ruled that a lower court erred when it denied a motion filed by the companies and the trade group for a preliminary injunction against a Maryland law. H.B. 1056, currently in effect, requires manufacturers to distribute discounted drugs to an unlimited number of pharmacies that contract with health providers under the 340B Drug Discount Program. They argued the law improperly forces drug companies to supply so-called contract pharmacies as part of the program, and that the law is illegal because it is preempted by federal law and also violates the U.S. Constitution.
The Download: NASA’s nuclear spacecraft and unveiling our AI 10
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.
NASA is building the first nuclear reactor-powered interplanetary spacecraft. How will it work?
Just before Artemis II began its historic slingshot around the moon, NASA revealed an even grander space travel plan. By the end of 2028, the agency aims to fly a nuclear reactor-powered interplanetary spacecraft to Mars.
A successful mission would herald a new era in spaceflight—and might just give the US the edge in the race against China. But the project remains shrouded in mystery.
MIT Technology Review picked the brains of nuclear power and propulsion experts to find out how the nuclear-powered spacecraft might work. Here’s what we discovered.
—Robin George Andrews
This story is part of MIT Technology Review Explains, our series untangling the complex, messy world of technology to help you understand what’s coming next. You can read more from the series here.
Coming soon: our 10 Things That Matter in AI Right Now
Each year, we compile our 10 Breakthrough Technologies list, featuring our educated predictions for which technologies will change the world. Our 2026 list, however, was harder to wrangle than normal. Why? We had so many worthy AI candidates we couldn’t fit them all in!
That got us thinking: what if we made an entirely new list all about AI? Before we knew it, we had the beginnings of what we’re calling 10 Things That Matter in AI Right Now.
On April 21, we’ll unveil the list on stage at our signature AI conference, EmTech AI, and then publish it online later that day. If you want to be among the first to see it, join us at EmTech AI or become a subscriber to livestream the announcement.
Find out more about the list’s methodology and aims here.
—Niall Firth & Amy Nordrum
MIT Technology Review Narrated: this company is developing gene therapies for muscle growth, erectile dysfunction, and “radical longevity”
In January, a handful of volunteers were injected with two experimental gene therapies as part of an unusual clinical trial. Its long-term goal? To achieve radical human life extension.
The therapies are designed to support muscle growth. The company behind them, Unlimited Bio, also plans to trial similar therapies in the scalp (for baldness) and penis (for erectile dysfunction). But some experts are concerned about the plans.
Find out why the trial has divided opinion.
—Jessica Hamzelou
This is our latest story to be turned into an MIT Technology Review Narrated podcast, which we publish each week on Spotify and Apple Podcasts. Just navigate to MIT Technology Review Narrated on either platform, and follow us to get all our new content as it’s released.
The must-reads
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 Google, Microsoft, and Meta track users even when they opt out
According to an independent audit, they may be racking up billions in fines. (404 Media)
+ How our digital devices put our privacy at risk. (Ars Technica)
+ Privacy’s next frontier is AI “memories.” (MIT Technology Review)
2 OpenAI has a new cybersecurity model—and strategy
GPT-5.4-Cyber is designed specifically for defensive cybersecurity work. (Reuters $)
+ OpenAI has joined Anthropic in focusing on cybersecurity recently. (Wired $)
+ Like Anthopic, its latest model is only available to verified testers. (NYT $)
+ AI is already making online crimes easier. It could get much worse. (MIT Technology Review)
3 Amazon is buying satellite firm Globalstar in a bid to rival Starlink
The $11.6 billion deal targets the lucrative satellite internet market. (WSJ $)
+ Apple has chosen Amazon satellites for iPhone. (Ars Technica)
4 What it’s like to live with an experimental brain implant
Early BCI users explain what the technology gives—and takes. (IEEE)
+ A patient with Neuralink got a boost from generative AI. (MIT Technology Review)
5 Dozens of AI disease-prediction models were trained on dubious data
A few might already have been used on patients. (Nature)
6 Uber is breaking from its gig economy model to avoid robotaxi disruption
It’s spending $10 billion to buy thousands of autonomous vehicles. (FT $)
7 xAI is being sued over data center pollution
Musk’s AI venture stands accused by the NAACP of violating the Clean Air Act. (Engadget)
+ No one wants a data center in their backyard. (MIT Technology Review)
8 Apple could win the AI race without running
It may reap the rewards of everyone else’s spending. (Axios)
9 How 4chan set a precedent for AI’s reasoning abilities
The notorious forum tested a feature called “chain of thought.” (The Atlantic $)
10 The surprising emotional toll of wearing Meta’s AI sunglasses
Their shortcomings are making users sad. (NYT $)
Quote of the day
“Everything got a whole lot worse once they rolled out AI.”
—A copywriter tells the Guardian that they’re drowning in “workslop” — AI-generated work that seems polished but has major flaws
One More Thing
GETTY IMAGES
How refrigeration ruined fresh food
Bananas may not be chilled in the grocery store, but they’re the ultimate refrigerated fruit. It’s only thanks to a network of thermal control that they’ve become a global commodity. And that salad bag on the shelf? It’s not just a bag but a highly engineered respiratory apparatus.
According to Nicola Twilley—a contributor to the New Yorker and cohost of the podcast Gastropod—refrigeration has wrecked our food system. Thankfully, there are promising alternative preservation methods.
Read the full story on her research.
—Allison Arieff
We can still have nice things
A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line.)
+ Spotify only shows 10 popular songs per artist. This tool lists them all.
+ These GIF animations are mesmerizing loops of nostalgia.
+ This site beautifully visualizes Curiosity’s 13 years on Mars.
+ A retro-futurist designer has turned a NES console into a working synthesizer.