Bible bytes: The Pope and Anthropic co-founder join forces on AI ethics
Copyright AP Photo/Andrew Medichini
Pope Leo's first major document will confront the age of AI.
More than a billion people follow it — and according to its believers, it is all-knowing. But it isn't a religion, it's artificial intelligence.
The use of the technology, though, is becoming a concern for many, with AI psychosis, cybersecurity, and the vast amounts of energy it needs to function.
Pope Leo XIV is one of those voicing their worries, and in the first major text of his papacy, he will address the impact of AI on humanity.
Pope Leo will present a document, known as an encyclical, at the Vatican on 25 May. The co-founder of AI company Anthropic, Christopher Olah, will attend the event.
Anthropic has billed itself as the AI company that puts safety and risk mitigation at the forefront of its research. But in February, the Trump administration ordered all US agencies to stop using Anthropic’s artificial intelligence technology and imposed other major penalties for refusing to allow the U.S. military unrestricted use of its AI technology.
Anthropic is currently suing the administration, which it has accused of retaliating against it illegally because it attempts to impose limits on how its AI technology can be deployed.
The encyclical will address “the protection of the human person in the age of artificial intelligence”, the Vatican said on Monday.
The encyclical is an official letter written by the Pope to guide bishops and practitioners.
The Pontiff has made AI a priority of his young pontificate, is greatly concerned about AI in warfare and has called for monitoring of how the technology is used.
He approved the creation of a new Vatican commission on artificial intelligence on 16 May 16.
The commission's role is to coordinate AI-related activities across Vatican institutions, sharing information, aligning on projects, and setting internal policies for AI use within the Holy See.
It draws from seven Vatican bodies, among them the Dicastery for the Doctrine of the Faith, the Pontifical Academy for Life, and the Pontifical Academy of Sciences.
More than a billion people follow it — and according to its believers, it is all-knowing. But it isn't a religion, it's artificial intelligence.
The use of the technology, though, is becoming a concern for many, with AI psychosis, cybersecurity, and the vast amounts of energy it needs to function.
Pope Leo XIV is one of those voicing their worries, and in the first major text of his papacy, he will address the impact of AI on humanity.
Pope Leo will present a document, known as an encyclical, at the Vatican on 25 May. The co-founder of AI company Anthropic, Christopher Olah, will attend the event.
Anthropic has billed itself as the AI company that puts safety and risk mitigation at the forefront of its research. But in February, the Trump administration ordered all US agencies to stop using Anthropic’s artificial intelligence technology and imposed other major penalties for refusing to allow the U.S. military unrestricted use of its AI technology.
Anthropic is currently suing the administration, which it has accused of retaliating against it illegally because it attempts to impose limits on how its AI technology can be deployed.
The encyclical will address “the protection of the human person in the age of artificial intelligence”, the Vatican said on Monday.
The encyclical is an official letter written by the Pope to guide bishops and practitioners.
The Pontiff has made AI a priority of his young pontificate, is greatly concerned about AI in warfare and has called for monitoring of how the technology is used.
He approved the creation of a new Vatican commission on artificial intelligence on 16 May 16.
The commission's role is to coordinate AI-related activities across Vatican institutions, sharing information, aligning on projects, and setting internal policies for AI use within the Holy See.
It draws from seven Vatican bodies, among them the Dicastery for the Doctrine of the Faith, the Pontifical Academy for Life, and the Pontifical Academy of Sciences.
The encyclical
The commission's creation sets the stage for Pope Leo's first encyclical, expected to tackle AI through the framework of Catholic social teaching — touching on labour rights, justice, and human dignity. The document, reportedly titled Magnifica Humanitas ("Magnificent Humanity"), is said to focus specifically on how AI is reshaping individuals and the workplace.
This isn't the Vatican's first foray into AI. Pope Francis spoke to the G7 on AI ethics in June 2024, and Vatican officials had long held private conversations with executives from Google, Microsoft, and Cisco on the same subject.
The Catholic Church's internal AI guidelines took effect on January 1, 2025, mandating disclosure of AI-generated content, prohibiting AI uses that run counter to the Church's mission, and creating a five-member compliance body to enforce the rules.
The commission's creation sets the stage for Pope Leo's first encyclical, expected to tackle AI through the framework of Catholic social teaching — touching on labour rights, justice, and human dignity. The document, reportedly titled Magnifica Humanitas ("Magnificent Humanity"), is said to focus specifically on how AI is reshaping individuals and the workplace.
This isn't the Vatican's first foray into AI. Pope Francis spoke to the G7 on AI ethics in June 2024, and Vatican officials had long held private conversations with executives from Google, Microsoft, and Cisco on the same subject.
The Catholic Church's internal AI guidelines took effect on January 1, 2025, mandating disclosure of AI-generated content, prohibiting AI uses that run counter to the Church's mission, and creating a five-member compliance body to enforce the rules.
Study maps how ‘Big AI’ influences AI laws and oversight
University of Edinburgh
Artificial intelligence (AI) companies influence policy and regulation using similar techniques to Big Tobacco, Big Pharma and Big Oil, according to a new study.
The influence methods that Big AI - companies that have developed and deployed large-scale AI technologies - have been mapped based on news articles around large global AI-focused events.
Researchers from the University of Edinburgh, Trinity College Dublin, TU Delft and Carnegie Mellon University, analysed news articles for evidence of methods used to control the narrative and influence policy measures related to AI.
They identified 27 established patterns of ‘corporate capture’ - a process by which regulation and public bodies come to act in the interest of corporations rather than people.
The researchers analysed 100 news stories published around four global AI events between 2023 and 2025 - the EU AI Act trilogues and the global AI summits in the UK, South Korea and France - and found 249 cases fitting capture patterns.
Of the mechanisms deployed, one of the most prevalent was ‘narrative capture’, which the team describe as attempts to influence the position or decisions of public officials and regulations.
The dominant narratives were around how “regulation stifles innovation” and “red tape”, whereby regulation is first portrayed as unnecessary, excessive or obsolete, setting the stage for later calls explicitly advocating for ‘deregulation’.
One of the other prevalent capture mechanisms used is what the team refers to as ‘elusion of law’, which relates to violations and contentious interpretations of antitrust, privacy, copyright and labour laws.
The research suggests that Big AI has undermined and resisted regulation, oversight and enforcement in a variety of ways, such as lobbying and retaliation against whistleblowers, researchers and lawmakers.
It was also found that in some cases the AI industry has benefited from a 'revolving door' model where former policymakers go on to advise or take employment with major AI companies.
There are also many examples of Big AI making significant donations to political parties and public officials owning equity in regulated companies, experts say.
The team highlight lessons to be learned from adjacent movements in similar industries such as Big Tobacco, Pharma and Oil on some of the tactics used to prevent capture.
These include calls for separation between public and private interests and binding rules for government-industry interactions to manage conflicts of interest.
The findings have been peer reviewed and will be presented at the ACM Conference on Fairness, Accountability, and Transparency in June 2026.
Dr Zeerak Talat, Chancellor’s Fellow at University of Edinburgh’s School of Informatics, said: “It’s remarkable how the findings relate to common experiences of companies having greater influence over democratic processes than people,” they further continue “While we cannot draw a causal relationship between attempts at corporate capture and the disenfranchisement of citizens, the former certainly seems to hint at the latter.”
Dr Abeba Birhane, Director of Trinity College Dublin’s AI Accountability Lab, said “In addition to ‘narrative capture’ and the violations and contentious interpretations of antitrust, privacy, copyright and labour laws that were most recurrent, we also found that Big AI frequently uses the notion that ‘regulation stifles innovation’ and that ‘red tape can stymy national interest’ to rationalise their control of the overall narrative.”
University of Edinburgh
Artificial intelligence (AI) companies influence policy and regulation using similar techniques to Big Tobacco, Big Pharma and Big Oil, according to a new study.
The influence methods that Big AI - companies that have developed and deployed large-scale AI technologies - have been mapped based on news articles around large global AI-focused events.
Researchers from the University of Edinburgh, Trinity College Dublin, TU Delft and Carnegie Mellon University, analysed news articles for evidence of methods used to control the narrative and influence policy measures related to AI.
They identified 27 established patterns of ‘corporate capture’ - a process by which regulation and public bodies come to act in the interest of corporations rather than people.
The researchers analysed 100 news stories published around four global AI events between 2023 and 2025 - the EU AI Act trilogues and the global AI summits in the UK, South Korea and France - and found 249 cases fitting capture patterns.
Of the mechanisms deployed, one of the most prevalent was ‘narrative capture’, which the team describe as attempts to influence the position or decisions of public officials and regulations.
The dominant narratives were around how “regulation stifles innovation” and “red tape”, whereby regulation is first portrayed as unnecessary, excessive or obsolete, setting the stage for later calls explicitly advocating for ‘deregulation’.
One of the other prevalent capture mechanisms used is what the team refers to as ‘elusion of law’, which relates to violations and contentious interpretations of antitrust, privacy, copyright and labour laws.
The research suggests that Big AI has undermined and resisted regulation, oversight and enforcement in a variety of ways, such as lobbying and retaliation against whistleblowers, researchers and lawmakers.
It was also found that in some cases the AI industry has benefited from a 'revolving door' model where former policymakers go on to advise or take employment with major AI companies.
There are also many examples of Big AI making significant donations to political parties and public officials owning equity in regulated companies, experts say.
The team highlight lessons to be learned from adjacent movements in similar industries such as Big Tobacco, Pharma and Oil on some of the tactics used to prevent capture.
These include calls for separation between public and private interests and binding rules for government-industry interactions to manage conflicts of interest.
The findings have been peer reviewed and will be presented at the ACM Conference on Fairness, Accountability, and Transparency in June 2026.
Dr Zeerak Talat, Chancellor’s Fellow at University of Edinburgh’s School of Informatics, said: “It’s remarkable how the findings relate to common experiences of companies having greater influence over democratic processes than people,” they further continue “While we cannot draw a causal relationship between attempts at corporate capture and the disenfranchisement of citizens, the former certainly seems to hint at the latter.”
Dr Abeba Birhane, Director of Trinity College Dublin’s AI Accountability Lab, said “In addition to ‘narrative capture’ and the violations and contentious interpretations of antitrust, privacy, copyright and labour laws that were most recurrent, we also found that Big AI frequently uses the notion that ‘regulation stifles innovation’ and that ‘red tape can stymy national interest’ to rationalise their control of the overall narrative.”
DOI
Method of Research
Literature review
Literature review
Subject of Research
Not applicable
Not applicable
Article Title
Big AI's Regulatory Capture: Mapping Industry Interference and Government Complicity
Big AI's Regulatory Capture: Mapping Industry Interference and Government Complicity
Article Publication Date
17-May-2026
17-May-2026
Policing plagiarism of ideas in generative AI-assisted research writing
Commentary: Onus is on researchers to use GenAI responsibly and ensure integrity, proper attribution
- Plagiarism of ideas harms the research environment by eroding trust among scientists
- Beyond researchers, plagiarism is an ethical concern for students, professionals in law, business and medicine
- ‘Checking AI output is still the simple and only way to ensure content is correct and reliable’
CHICAGO — As more people — including researchers — use generative artificial intelligence (GenAI) in their writing, it’s becoming increasingly important to define what plagiarism looks like and how to police it.
A new commentary written by researchers at Northwestern University and the National Institutes of Health (NIH) that will publish May 18 in Nature Machine Intelligence weighs various options on how to define plagiarism in research manuscript writing in the ever-evolving world of GenAI.
The commentary argues plagiarism in manuscript writing harms the research environment by eroding trust among scientists, misrepresenting the origin and authenticity of scholarly work, and discouraging innovation and original inquiry.
Currently, most plagiarism policies address two types of plagiarism: plagiarism of original works, such as text or verbatim plagiarism, and plagiarism of ideas. Since GenAI tools can easily rephrase text, verbatim plagiarism is becoming less of an issue. But plagiarism of ideas, which is essentially a form of intellectual theft, is still an important concern because a GenAI tool may plagiarize underlying ideas without giving appropriate credit.
“It is fine and in fact helpful to use GenAI to increase the readability of writing and bounce ideas back and forth, but we know these tools frequently make mistakes of fact and accuracy and have enormous social and environmental impacts,” said corresponding author Mohammad Hosseini, assistant professor of preventive medicine in the division of biostatistics and informatics at Northwestern University Feinberg School of Medicine. “Checking AI output is still the simple and only way to ensure content is correct and reliable.”
Because plagiarism of ideas is difficult to detect and enforce, the commentary recommends that definitions of research misconduct — which includes plagiarism as well as data fabrication and falsification — be revised to include that misconduct may be committed by a person when using GenAI tools.
“If a person using GenAI tools does not do their own background research and carefully review the GenAI’s output, they may not be aware that the tool has plagiarized,” Hosseini said. “By revising the definition, we’re hoping to make it clear that those who use GenAI tools are responsible for avoiding plagiarism, which will ultimately promote more responsible use of GenAI tools.”
Enforcement of plagiarism goes beyond researchers
When universities or funders conclude that a researcher has committed research misconduct, they may impose sanctions with serious adverse career consequences, including retractions, loss of current funding or debarment from future grants, termination of employment or revocation of academic degrees. But the study authors said this discussion goes beyond writing in the research world.
“Non-researchers should also use GenAI in responsible ways,” Hosseini said. “Plagiarism is an ethical and legal concern not just for researchers but also for students and those working in various professions, such as law, business and medicine.”
The commentary is titled, “Plagiarism of ideas in the age of generative artificial intelligence.” David Resnick, a senior bioethicist at the NIH, is a co-author.
Journal
Nature Machine Intelligence
Article Title
Plagiarism of ideas in the age of generative artificial intelligence
Article Publication Date
18-May-2026
New AI closes data gaps and shows how extreme weather emerges on Earth
ETH Zurich
image:
The Earth System Foundation Model can also complete missing data in satellite images – as in the case of the MODIS sensor, which provides global Earth observation data. The satellite image shown here is incomplete.
view moreCredit: Firat Ozdemir / SDSC
The impacts were severe: Within a very short time, tropical storm Doksuri intensified into a super typhoon in July 2023. Exceptionally strong winds tore roofs from houses along the coasts of China and the Philippines, trees were uprooted, and torrential rain flooded streets and residential areas. In many places, everyday life came to a temporary halt.
Extreme events such as Super Typhoon Doksuri are particularly difficult for weather and climate models to predict, as they arise from complex interactions between the atmosphere, the land surface and the water cycle.
Researchers from the ETH Domain have now introduced a new artificial intelligence (AI) model that has learned these interactions and feedback autonomously – without human guidance – and, compared to previous AI models, more precisely captures how air, land and water interact on Earth.
AI understands Earth system’s key connections
The new Earth System Foundation Model (ESFM) does not treat atmospheric and hydrological (i.e. water-related) processes in isolation, but rather represents them as part of an interconnected Earth system.
“Previous AI weather models have often focussed primarily on the atmosphere. Our model, by contrast, deliberately links atmospheric weather data with hydrological and land-based data. On this basis, the AI identifies key patterns, trends and relationships within the Earth’s weather system and uses them to generate forecasts, even when important data is missing,” explains Fanny Lehmann, mathematician, ETH AI Center Postdoctoral Fellow, and member of the team that developed the new model.
“The true strength of our model lies in its ability to learn the interactions that are crucial for weather from different data sources. This allows ESFM to integrate very different and hard‑to‑compare data types and to analyse them jointly for the first time.”
The researchers tested their model using Super Typhoon Doksuri as a case study. This tropical storm was not part of the training data. Even so, ESFM predicted wind strength with remarkable accuracy over several days and simultaneously captured realistically where the storm was, how quickly it moved, and how it expanded in space. This demonstrated how effectively the new model can jointly process very large, complex and heterogeneous datasets.
Learning from incomplete and heterogeneous data
The integrative approach of ESFM addresses a need in climate and environmental sciences. In research practice, data often varies considerably: some comes from satellite imagery, some from weather balloons, ground‑based stations, or other sensors. This data ranges from very fine-grained, short‑term measurements to large‑scale, long‑term observations.
Data types also differ markedly. While satellite imagery and climate models provide data in the form of large‑scale raster maps, ground stations or wells record key variables such as temperature, air pressure, wind speed or water levels at specific locations and at defined points in time.
To integrate these different types of environmental data, ESFM follows a multi‑stage approach: rather than forcing all data types into a single format from the outset, it initially treats them separately, depending on their type – whether satellite or station data – and tags them with information on when and where they were measured.
This approach enables the combination of very different data within a common spatial and temporal framework, while preserving its specific information. On this basis, the model learns the typical, recurring process chains and fundamental relationships within the Earth system.
Maintains performance despite missing data
“Earlier AI models for weather forecasting – unlike ESFM – were often trained on a single type of data or on a few datasets of a similarly formatted nature,” explains Firat Ozdemir, lead developer of the ESFM team and Senior Data Scientist at the joint Swiss Data Science Center of ETH Zurich and EPFL. “Their performance often declines when working with highly heterogeneous or incomplete data. ESFM addresses this challenge by integrating multi-source data and filling data gaps much more efficiently.”
“ESFM is neither a classical climate model nor a weather forecasting or specialised storm‑warning model; rather, it belongs to a distinct category of models that can serve as a flexible foundation for a wide range of tasks in climate and weather research,” says Sebastian Schemm, atmospheric scientist and professor at the University of Cambridge, formerly at ETH Zurich.
“Its advantage lies in a kind of learned systemic understanding that enables it to produce plausible predictions in many cases, even when data is incomplete or patchy.”
Designed to bridge data gaps intelligently
Such data gaps significantly hindered previous AI models in analysing and predicting complex weather and water phenomena. However, in research practice, it is not uncommon for individual measurements to be lost or compromised due to weather conditions or technical issues. Measurement networks, too, often contain gaps, as monitoring stations are unevenly distributed.
ESFM, by contrast, is specifically designed to cope with missing data and to internally reconstruct incomplete observations, such as patchy satellite images. After training, the model succeeds in generating forecasts from satellite observations in which only around 3 percent of the pixels are available.
The researchers, including Benedikt Soja, Professor of Space Geodesy at ETH Zurich, showed that their model can reliably fill data gaps both in weather station data and in the long-term global ERA5 dataset. On this basis, it is able to generate plausible forecasts of weather conditions.
Building on many learned examples of how the atmosphere, land and water are interconnected, ESFM can plausibly complete patchy satellite images with information such as temperature, humidity, soil type, whether an area is land or sea, and topography.
The model systematically embeds this information within the processes that link, for example, rainfall, soil moisture and groundwater, thereby helping to improve the understanding of droughts and potentially making them easier to predict.
ESFM infers missing measurement data by relating data gaps to other available data sources and to patterns it has learned from similar situations in neighbouring regions, from related variables, and from past observations.
Learning is more than repetition
“Through training on very different types of data, models such as ESFM acquire a form of fundamental knowledge and can therefore flexibly solve a wide range of tasks. In AI research, they are referred to as foundation models,” says Torsten Hoefler, Professor of Computer Science at ETH Zurich, who also serves as Chief AI Architect at the Swiss National Supercomputing Centre (CSCS) in Lugano, where he oversees research on new AI approaches (see box).
Like all foundation models, ESFM can be used for a range of tasks and can also be adapted to specific applications through a process known as finetuning. The team’s research shows that ESFM applies fundamental physical principles consistently and reliably – even when addressing new physical or weather‑related questions or working with variables for which it was not explicitly trained.
In the future, ESFM or especially finetuned versions have the potential to provide reliable forecasts of weather and water processes. “We intend to leverage the model’s representational power across diverse domains such as agriculture, biodiversity and hydrology,” says Mathieu Salzmann, Senior Scientist at EPFL and Deputy Chief Data Scientist at the Swiss Data Science Center (SDSC).
Swiss AI Initiative, ICAIN and download
ESFM was developed within the Weather and Climate Foundation Models project, which is part of the Swiss AI Initiative. The project also includes ETH Zurich mathematics professor Siddhartha Mishra. Within this framework, researchers from ETH Zurich, EPFL and other partners are developing foundation models addressing key challenges in Switzerland.
ESFM is also supported by the International Computation and AI Network (ICAIN) at ETH Zurich. The network promotes international AI collaboration and works to ensure that such foundation models can be used in the Global South. Within the ESFM project, ICAIN helps identify partners in data‑sparse regions to enable finetuning of the model with local data.
ESFM is freely available on the AI platform Hugging Face and in the Git repository.
Reference
Ozdemir, F., Cheng, Y., Mohebi, S., Lehmann, F., Adamov, S., Trentini, L., Huang, L., Lingsch, L., Zhang, Z., Fuhrer, O., Soja, B., Mishra, S., Hoefler, T., Schemm, S., and Salzmann, M.: ESFM – A foundation model framework for heterogeneous data integration. EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18011. DOI:10.5194/egusphere-egu26-18011.
This is the fully reconstructed image, with global coverage restored by the AI.
Credit
Firat Ozdemir / SDSC
Article Title
ESFM - A foundation model framework for heterogeneous data integration
No AI authors, no fake citations: international consensus sets rules for AI-assisted medical writing
A ghostwriter that never sleeps but sometimes lies. That is the double-edged reality of generative artificial intelligence (AI) in medical research. A new global consensus delivers the first practical playbook for using tools like ChatGPT responsibly. The work tackles the biggest fears head-on: fake references, data leaks, and machines masquerading as authors. Instead of banning AI, the guidelines show exactly where it helps—polishing clunky English, for example—and where it is strictly forbidden, such as inventing experimental results or creating research images. The goal is to protect scientific integrity without forcing researchers back to typewriters.
Most major publishers now have artificial intelligence (AI) policies, but they are a mess of contradictions. One journal demands every prompt and software version; another just says "be honest." This inconsistency leaves scientists in a grey zone, unsure if using spell-checker AI crosses an ethical line. Worse, large language models are notorious "hallucinators," confidently generating plausible but completely fake citations or scientific nonsense. There is also the privacy nightmare: feeding unpublished patient data into a public chatbot. Because of these risks, a clear, unified, and practical framework is urgently needed to guide researchers without punishing them for using modern tools.
Researchers led by experts from The First Affiliated Hospital of Shenzhen University, China, working with a multidisciplinary international panel that included contributors from universities, hospitals, and editorial institutions in China, Italy, Japan, Canada, Australia, and other countries, publishes (DOI: 10.1016/j.rerere.2026.02.001) the new guidance on March 10, 2026, in the journal Regenesis Repair Rehabilitation. The team first audited AI rules from 15 major publishers, including Elsevier and the JAMA Network. They found broad agreement on principles but chaos in practice. Using a structured Delphi process, they built a consensus that fits real-world research workflows, resulting in a simple, section-by-section checklist for authors to disclose exactly how they used AI.
The new rules draw a hard line in the sand. AI can fix your grammar, but it cannot touch your data. The guidelines strictly prohibit using generative AI to create or manipulate primary research images, including microscopy, gels, and flow cytometry plots. For references, the message is clear: do not let AI generate them. The tools are too prone to fabricating authors, journal names, and DOIs. However, AI gets a green light for supportive tasks, like summarizing a discussion or refining an abstract, as long as a human verifies every claim.
A standout feature is the section-specific guidance. In the Methods section, AI can improve readability but cannot invent missing steps. In the Results section, the text can be polished, but the numbers themselves must come from genuine experiments. For non-data visuals, like workflow diagrams, AI is allowed but must be fully disclosed. The authors even put their own advice into practice, using Gemini and ChatGPT during writing—but only for language and structure, never for core science. The final takeaway is simple: AI is a powerful assistant, but the human author remains the sole, accountable author.
"We are not the AI police, and we are not telling people to throw away their laptops," the authors said. "The worst outcome would be researchers secretly using AI anyway, afraid to admit it." They explained that the real danger is not AI itself, but invisibility. "If a chatbot rewrites your discussion or suggests a citation, that is fine—just put it in a table. The moment you claim AI-generated text as your own original thought, or a fabricated reference as real, you have crossed into misconduct." They added: "Transparency turns a black box into a simple tool. And that is all this should be."
For researchers, the consensus removes guesswork. They now have a checklist they can hand to collaborators or include in supplementary files. For journals, the standard offers a ready-made audit trail to streamline peer review. The guidelines are especially valuable for non-native English speakers, who can confidently use AI for language help without ethical fear. As AI evolves, the authors commit to updating the rules regularly. Ultimately, this work shifts the conversation from "is AI cheating?" to "how do we use AI well?" — protecting science from silent errors while embracing the efficiency that modern writing tools provide.
###
References
DOI
Original Source URL
https://doi.org/10.1016/j.rerere.2026.02.001
About Regenesis Repair Rehabilitation
Regenesis Repair Rehabilitation publishes rigorously peer-reviewed high-quality research articles, concise reviews, clinical studies, and short communications on topics relevant to the tissue/organ regeneration, repair and rehabilitation, particularly the skin tissue and its appendages. The journal covers the stem cell areas concerned with investigations about stemness maintenance and differentiation, organoids for grafting, and materials/biomaterials for injured tissue protection, healing promotion, and regeneration, 3D printing for tissue reconstruction, and novel techniques, methods, and progresses in the surgical and rehabilitation areas. The submission of manuscripts conveying original findings or thoughtful reviews concerning all the above fields of endeavor, which potentially contribute to Translational Medicine, are encouraged.
Journal
Regenesis Repair Rehabilitation
Subject of Research
Not applicable
Article Title
Consensus on the application of generative artificial intelligence in medical manuscript writing
AI adoption boosts educational effectiveness in emerging higher education institutions
New study finds digital literacy strengthens the impact of AI adoption on teaching, learning, and institutional performance in Pakistani universities
Bentham Science Publishers
As artificial intelligence rapidly reshapes higher education worldwide, a new study highlights the growing importance of AI adoption and digital literacy in improving educational effectiveness in emerging economies. Researchers investigating Pakistani higher education institutions found that AI adoption significantly enhances teaching quality, learning efficiency, academic decision-making, and institutional performance, particularly when supported by strong digital literacy skills.
The study examined the determinants of AI adoption within higher education institutions and explored how AI-driven technologies influence educational effectiveness. Researchers found that institutions integrating AI-powered tools for learning management, academic analytics, administrative automation, and personalized education were better positioned to improve student engagement, operational efficiency, and overall educational outcomes.
Importantly, the findings revealed that digital literacy plays a significant moderating role in strengthening the positive relationship between AI adoption and educational effectiveness. Faculty members and students with stronger digital competencies were more capable of effectively utilizing AI systems, maximizing the benefits of intelligent educational technologies, and adapting to rapidly evolving digital learning environments.
The researchers emphasize that successful AI integration in higher education depends not only on technological infrastructure but also on institutional readiness, training, and human capability development. The study recommends that universities invest in digital literacy programs, AI training initiatives, and supportive technology policies to ensure sustainable and inclusive educational transformation.
The findings contribute to growing global discussions on AI-driven educational innovation and provide important insights for policymakers and university administrators seeking to improve higher education quality in emerging economies through responsible AI adoption.
New study finds digital literacy strengthens the impact of AI adoption on teaching, learning, and institutional performance in Pakistani universities
Bentham Science Publishers
As artificial intelligence rapidly reshapes higher education worldwide, a new study highlights the growing importance of AI adoption and digital literacy in improving educational effectiveness in emerging economies. Researchers investigating Pakistani higher education institutions found that AI adoption significantly enhances teaching quality, learning efficiency, academic decision-making, and institutional performance, particularly when supported by strong digital literacy skills.
The study examined the determinants of AI adoption within higher education institutions and explored how AI-driven technologies influence educational effectiveness. Researchers found that institutions integrating AI-powered tools for learning management, academic analytics, administrative automation, and personalized education were better positioned to improve student engagement, operational efficiency, and overall educational outcomes.
Importantly, the findings revealed that digital literacy plays a significant moderating role in strengthening the positive relationship between AI adoption and educational effectiveness. Faculty members and students with stronger digital competencies were more capable of effectively utilizing AI systems, maximizing the benefits of intelligent educational technologies, and adapting to rapidly evolving digital learning environments.
The researchers emphasize that successful AI integration in higher education depends not only on technological infrastructure but also on institutional readiness, training, and human capability development. The study recommends that universities invest in digital literacy programs, AI training initiatives, and supportive technology policies to ensure sustainable and inclusive educational transformation.
The findings contribute to growing global discussions on AI-driven educational innovation and provide important insights for policymakers and university administrators seeking to improve higher education quality in emerging economies through responsible AI adoption.
DOI
Method of Research
Survey
Survey
Subject of Research
Not applicable
Not applicable
Article Title
AI Adoption and Educational Effectiveness in Emerging Higher Education Institutions: The Moderating Role of Digital Literacy and Institutional Support
AI Adoption and Educational Effectiveness in Emerging Higher Education Institutions: The Moderating Role of Digital Literacy and Institutional Support
Academic publishes book on AI and business agility
New publication explores how AI and digital innovation can help organisations stay competitive in a rapidly changing economy
A University of East London academic has published a major new book examining how businesses can stay competitive in an era shaped by artificial intelligence, rapid technological change and economic disruption.
DigiTech Agility for Business Competitiveness and Innovation Imperatives explores how organisations can use technologies including AI, blockchain, cloud systems and data analytics to become more adaptable, resilient and innovative.
The book has been edited by Professor Nazrul Islam, of the Royal Docks School of Business and Law, and colleagues and brings together international perspectives on digital transformation, business strategy and innovation.
Covering topics ranging from AI governance to fintech, entrepreneurship and global digital competition, it offers practical frameworks and case studies designed to help organisations respond to disruption.
The publication argues that successful digital transformation is not simply about adopting new technologies. Instead, organisations must rethink leadership, workforce development and organisational culture if they are to compete effectively.
One of the book’s central themes is that agility is now essential for survival. Businesses face mounting pressure from automation, cybersecurity risks, changing consumer expectations and economic instability.
Professor Islam, Chair Professor of Business and Associate Director of the Centre of FinTech, said:
“This book was written to help organisations understand that digital transformation is ultimately about people, strategy and responsible leadership, not just technology.
“AI and emerging technologies have enormous potential to improve productivity, innovation and decision-making, but businesses must also develop the skills and capability to use these tools responsibly. I hope the book provides practical guidance for leaders, policymakers and researchers who are shaping the future of business in the digital age.”
DigiTech Agility for Business Competitiveness and Innovation Imperatives is published by IGI Global Scientific Publishing and is available from the publisher’s website and major academic book suppliers.
AI-enhanced project-based learning bridges STEM educational gaps in Africa
New research offers a scalable model for transitioning to competency-based learning in resource-limited classrooms using low-cost AI tools
A generative artificial intelligence (AI) framework is proving to be a transformative enabler for science, technology, engineering, and mathematics (STEM) education in Sub-Saharan Africa. In a study published online on May 5, 2026, in ECNU Review of Education, a research team led by Sanura Jaya and Rozniza Zaharudin from Universiti Sains Malaysia investigated how AI-enhanced project-based learning (PBL) can empower educators in resource-constrained environments. The research highlights a critical shift from traditional content-centric teaching to competency-based learning (CBL) by leveraging affordable, smartphone-based technologies.
Educational systems worldwide are increasingly adopting AI, yet low-resource classrooms often remain at the margins of this digital transformation. This study addresses the persistent gap between policy aspirations and classroom realities in Africa, where infrastructural and capacity limitations often hinder the adoption of emerging technologies. By focusing on "leapfrogging technology," the researchers demonstrated that AI can compensate for deficits in laboratory equipment and teacher training.
The qualitative research involved ten STEM educators from Nigeria, Botswana, Ghana, Namibia, and Sierra Leone. These participants engaged in a hands-on capacity-building workshop that integrated speech-to-text-to-image (STTI) generation, smartphone-based block coding via the Magnetcode application, and circuit simulations. Using Kolb's experiential learning theory as a guide, the study followed educators through cycles of concrete experience, reflection, and active experimentation.
The researchers identified five major outcomes that underscore the potential of AI in under-resourced schools. First, AI STTI tools significantly enhanced the visualization of abstract science concepts, acting as a cognitive scaffold for learners facing language barriers or limited exposure to scientific imagery. Participants noted that converting verbal prompts into immediate visual feedback made learning biological and physical concepts more accessible.
Secondly, the study found that smartphone-based block coding increased digital inclusion. While laptops are often scarce, the high penetration of mobile devices in Africa allows learners in rural schools to develop computational thinking (CT) skills without expensive hardware. The Magnetcode application provided a simplified interface that minimized programming syntax barriers, allowing educators to focus on logic and problem-solving.
A third key finding was the effectiveness of simulation tools as a cost-effective bridge to physical hardware. Simulations allowed educators to experiment with and debug electronic circuits in a low-risk virtual environment before transitioning to physical microcontrollers. This "simulation-first" strategy builds confidence and prevents the accidental damage of costly equipment, which is vital in schools with limited lab resources.
Beyond technical skills, the study revealed a significant pedagogical shift. Educators reported transitioning from teacher-centered delivery toward a facilitative role, focusing on student-driven inquiry and creativity. This evolution is essential for embedding CT as a core competency in twenty-first-century STEM education. One participant reflected that the training built the necessary confidence to integrate problem-solving tasks into daily lessons rather than relying solely on rote content delivery.
Finally, the participating educators articulated concrete plans for classroom integration, such as establishing extracurricular AI clubs and using recycled materials for coding projects. These strategies demonstrate proactive agency in adapting AI tools to local contexts despite systemic barriers like restrictive device policies. The researchers emphasize that these learning outcomes were driven by pedagogical design, with AI serving as an enabling component rather than a standalone solution.
"This study offers scalable and context-responsive models for advancing inclusive, technology-enhanced STEM education globally," Sanura Jaya et al. conclude. They recommend that AI and CT be systematically embedded into STEM curricula through inquiry-driven modules that emphasize both simulation and physical prototyping. By aligning policy and practice, schools can create future-ready learning environments that promote equity and technological empowerment.
Journal
ECNU Review of Education
Method of Research
Case study
Subject of Research
People
Article Title
Bridging Educational Gaps in Low-Resource Classrooms: AI-Enhanced Project-Based Learning for STEM Educators in Africa
Article Publication Date
15-May-2026
Therapy at your fingertips: new Reichman University study led by Prof. Anat Shoshani of the Baruch Ivcher School of Psychology finds AI could transform mental health care
image:
Prof. Anat Shoshani of the Baruch Ivcher School of Psychology, Reichman University
view moreCredit: Gilad Kavalerchik
A new study from Reichman University, published in the prestigious journal JAMA Network Open, has found that an AI-based conversational support platform can significantly reduce symptoms of anxiety and depression, form a meaningful therapeutic alliance with users, and provide round-the-clock emotional support.
Led by Prof. Anat Shoshani of the Baruch Ivcher School of Psychology at Reichman University, and conducted in collaboration with the Kai.ai platform, the study explored whether AI systems could offer an innovative response to the global mental health crisis. Approximately 1,000 Israeli students experiencing emotional distress, anxiety, and depression participated in the study. They were assigned to one of three groups: traditional in-person group therapy with psychologists, a waitlist control group, or a group given access to “Kai,” an AI-powered emotional support platform operating within a popular messaging app and available to users at any time.
The study was conducted during a period of prolonged security tensions in Israel, which heightened the need for accessible and immediate mental health support. The “Kai” system is grounded in established scientific protocols, including CBT, ACT, DBT, mindfulness, and positive psychology. It is capable of sustaining ongoing conversations, recalling past interactions, identifying signs of distress, and offering real-time tools for emotional regulation, breathing exercises, and reflective writing.
The findings showed that users of the conversational AI platform experienced a significant reduction in symptoms of anxiety and depression compared to the waitlist group. In reducing anxiety, the system even outperformed human-led group therapy. Notably, approximately 58% of participants who initially presented with clinical levels of anxiety moved into the healthy range following the intervention. Among participants experiencing depression, nearly half reported substantial improvement.
One of the study’s most striking findings was the formation of a “therapeutic alliance” — a sense of trust, understanding, and emotional connection between users and the AI platform. Participants rated the system as empathetic, professional, and supportive at levels comparable to human therapists. Many also reported that they found it easier to open up to a digital system, free from concerns about judgment or embarrassment.
Prof. Anat Shoshani of Reichman University’s Baruch Ivcher School of Psychology explains: “Anxiety is highly situational. Clinical therapy is invaluable, but it is episodic. AI sits in the user’s pocket — on the bus, in the library, and during sleepless nights — providing continuous support precisely when it is needed. The goal is not to create machines that sound more human, but to build systems that help make our society more human. The therapy of the future will be a continuum of support, available where life actually happens.”
While many mental health apps experience high dropout rates, the study found that 61% of users continued to engage with the platform throughout the 12-week study, using it on average three days per week. According to the researchers, the sense of continuity, personalized availability, and the system’s ability to “remember” users and their personal context fostered a deeper connection that sustained engagement over time.
Notwithstanding these promising findings, the researchers stress that AI is not a substitute for human psychological treatment, particularly in complex conditions such as post-traumatic stress disorder (PTSD). The Kai.ai platform operates within a hybrid model that includes a human support team available 24/7. When the system detects signs of acute distress or risk, it triggers an alert to enable immediate human intervention.
The full study was published on April 14, 2026, in JAMA Network Open, under the title “Efficacy of a Conversational AI Agent for Psychiatric Symptoms and Digital Therapeutic Alliance: A Randomized Clinical Trial.”
Journal
JAMA Network Open
TEGNet: AI that freely designs thermoelectric devices
Innovating the development process by only requiring about 1/10,000 of the time conventionally needed for predicting performance
image:
Significant acceleration of prediction of thermoelectric device performance using AI model TEGNet (shortening the computational time to about 1/10,000 of the time conventionally needed)
view moreCredit: Takao Mori, National Institute for Materials Science
NIMS developed TEGNet (Thermoelectric Generator Neural Network), a neural network for designing thermoelectric generators by utilizing artificial intelligence (AI). TEGNet can predict performance of a power generator, a process which used to take enormous computational time with traditional simulation techniques, with only about 1/10,000 of the time conventionally needed, while maintaining over 99% accuracy. This technology significantly accelerates optimization from material development to device design, and is expected to be applied to waste heat recovery and stand-alone power supply for IoT sensors, for example. This research result was published in Nature at 11:00 U.S. Eastern Standard Time, April 15, 2026 (0:00 Japan Standard Time, April 16, 2026).
Background
Toward realizing a sustainable society, thermoelectric generation technology capable of generating power indefinitely merely by installing a device at any place with temperature differences is drawing attention. In order to improve performance of thermoelectric generators, not only material development, but also optimal design of dimensions and structure are indispensable. However, conventional numerical analysis (finite element method) had a problem of requiring computation to be repeated whenever conditions were changed, which caused heavy computational load and made large-scale and high-speed design exploration difficult.
Key Findings
In order to solve this problem, the research group developed TEGNet, an AI model capable of optimizing the design of thermoelectric generators at high speed. If material properties and element dimensions and conditions are input into TEGNet, TEGNet quickly predicts voltage and heat flow that are generated within the device, making it possible to estimate power generation output and conversion efficiency at high accuracy (Figure 1). The most notable feature of this technology is in its composability, which allows designers to freely combine independent TEGNet models that have been trained for each material, like blocks, based on the laws of physics. This approach can enable designers to speedily and exhaustively explore and optimize performance, even for a device with a complex structure that combines materials having different properties. To demonstrate this approach, the team optimized two types of device designs using Mg-Sb (magnesium-antimony) based materials, prototyped and evaluated those devices, and achieved high conversion efficiencies of up to 9.3% and 8.7% under practical temperature conditions.
Future Outlook
This research result proposes a next-generation design technique using AI as core technology, as opposed to conventional thermoelectric generator design dependent on numerical simulations. While many AI studies have focused on material-level optimization in recent years, the key feature of this research is that it directly targets device-level optimization. As a result, it becomes possible to sophisticate material design and device design in a complementary manner, and AI utilization is expected to develop further not only in the thermoelectric field, but in the entire energy field.
Other Information
- This project was conducted by a research team led by Takao Mori (Group Leader, Thermal Energy Materials Group, Nanomaterials Field, Research Center for Materials Nanoarchitectonics (MANA), NIMS). The work was supported by the Japan Science and Technology Agency (JST), JST-Mirai Program Large-Scale Type, technology theme: “Innovative thermoelectric conversion technologies for stand-alone power supplies for sensors” (Project Leader: Takao Mori).
- This research result was published online in Nature at 11:00 U.S. Eastern Standard Time, April 15, 2026 (0:00 Japan Standard Time, April 16, 2026).
Journal
Nature
Method of Research
Experimental study
Subject of Research
Not applicable
Article Title
Composable neural emulators accelerate thermoelectric generator design
Economic impact report examines the value of open biodata infrastructure
EMBL-EBI enables global productivity gains worth £11.8bn per year and powers AI-driven science and innovation at scale
European Molecular Biology Laboratory
Public infrastructures like roads and electricity are so essential to society that people almost take their value for granted. A new report by Frontier Economics, commissioned by EMBL’s European Bioinformatics Institute (EMBL-EBI), shows that open data resources have become a similarly essential infrastructure for life sciences worldwide.
EMBL-EBI is a global leader in the storage, analysis, and dissemination of large biological datasets across scales and disciplines. The open data resources EMBL-EBI manages support growing numbers of scientists and innovators in academia and industry.
This is the third independent economic assessment of these resources. First introduced in 2016 and updated in 2021, the report builds a unique decade-long picture of how life scientists use and value EMBL-EBI data resources. It shows a diversification of users and a tripling of the returns on research and development enabled by these resources. The findings are indicative of the broader value of open data in the life sciences.
“Most EMBL-EBI data resources exist thanks to global collaborations and joint funding,” said Jo McEntyre, Interim Director of EMBL-EBI. “No single institute or country can manage the scale of today’s biological data. This vital shared infrastructure enables breakthroughs across science, medicine, and biotechnology, but its continued impact relies on stable, long-term investment. We are incredibly grateful to our funders, collaborators, and users for their support and the trust they place in EMBL-EBI.”
Enabling productivity gains
The report combines anonymised web access data with a global survey of over 2,500 EMBL-EBI data resource users across the public and private sectors. It estimates that EMBL-EBI enables productivity gains worth £11.8bn per year, driven by average savings of 11 hours per user, per week.
By providing open access to vast volumes of high-quality, expertly curated data, EMBL-EBI resources reduce duplication of effort across the life sciences.
They remove the need for scientists to repeatedly generate the same datasets, freeing up time and funding to focus on new discoveries.
The report highlighted that 71% of respondents say EMBL-EBI enables work that would otherwise be impossible or require significant additional time and effort. This underlies the institute’s critical role in the global research ecosystem.
Powering AI-driven science and innovation at scale
The report found that EMBL-EBI data resources are catalysing a growing global research and innovation ecosystem. More than a third of survey respondents said they build new tools and databases on top of EMBL-EBI data resources, extending their value and usage across different disciplines.
This role is increasingly important in the era of AI-driven science, where access to high-quality training data is essential. Indeed, 42% of survey respondents said that EMBL-EBI data resources contribute to AI and machine learning model development.
A leading example of this is Google DeepMind’s AlphaFold 2, an AI-based system which accurately predicts the structure of proteins. This information is important for understanding the biology of disease, and for developing new drugs and treatments.
AlphaFold was trained on public data resources, including those managed by EMBL-EBI. Crucially, EMBL-EBI worked with Google DeepMind to make over 200 million protein structure predictions openly available to the global scientific community via the AlphaFold Database. The report studied the impact of the database and found that it had widened access to the algorithm’s predictions for a more diverse array of research fields, and likely increased the total volume of research.
“Estimating the value and socio-economic impacts of open data resources is not easy,” said Thomas Badger from Frontier Economics. “Our conservative calculations show that EMBL-EBI data resources deliver substantial and growing benefits to users and society. They support a global research community, enabling work that would otherwise not be possible, and save researchers a significant amount of time.”
Read the full report on the EMBL-EBI website to find out more.
EMBL's European Bioinformatics Institute (EMBL-EBI)
EMBL's European Bioinformatics Institute (EMBL-EBI) is a global leader in the storage, analysis and dissemination of large biological datasets. We help scientists realise the potential of big data by enhancing their ability to exploit complex information to make discoveries that benefit humankind.
We are at the forefront of computational biology research, with work spanning sequence analysis methods, multi-dimensional statistical analysis and data-driven biological discovery, from plant biology to mammalian development and disease.
We are part of EMBL and are located on the Wellcome Genome Campus, one of the world’s largest concentrations of scientific and technical expertise in genomics.
Website: www.ebi.ac.uk
AI-powered CPR coach outperforms 911 dispatchers in guiding bystander resuscitation
University of California - San Diego
image:
A new study in JAMA Internal Medicine found that ChatCPR outperformed 911 dispachers in guiding bystander CPR.
view moreCredit: Courtesy of John W. Ayers, UC San Diego Qualcomm Institute
A new study from scientists at the University of California San Diego in collaboration with the University of Pittsburgh School of Medicine, Johns Hopkins University, and other institutions, demonstrates that an artificial intelligence-powered CPR coaching agent can outperform 911 dispatchers in guiding bystanders through cardiopulmonary resuscitation.
The study, published in JAMA Internal Medicine, first benchmarks popular AI models on CPR instruction and then introduces ChatCPR, an open-source AI agent that scored 100% on guideline-based CPR checklists and significantly outperformed human dispatchers when tested against recordings from previous real 911 calls.
“If AI is going to earn its place in medicine, it should start by helping people save the person right in front of them,” said John W. Ayers, Ph.D., who is a UC San Diego Qualcomm Institute scientist, in addition to Head of AI at the UC San Diego Altman Clinical and Translational Research Institute, professor at the UC San Diego School of Medicine, and study coauthor.
“More than 350,000 Americans suffer out-of-hospital cardiac arrest each year, and survival sits at roughly 9%. Given that only 2% of Americans are certified to perform CPR, when someone collapses, they call 911 and wait. ChatCPR could change that and begin to save lives,” added Nimit Desai, B.S., a research affiliate at the UC San Diego Qualcomm Institute, a medical student at UC San Diego School of Medicine, and lead author of the study.
A First Step Toward AI-Assisted CPR
Adapting new technologies could help close the gap between cardiac arrest and initiation of CPR. In this proof-of-concept study, researchers used 12 real 911 calls to compare AI-generated CPR instructions with those provided by dispatchers.
“As new AI technologies emerge, we know people are going to start using them in real-world situations,” said Christopher M. Horvat, M.D., Director of Medical Emergency Response Teams at UPMC Children’s Hospital of Pittsburgh. “Our goal was to take a first step to understand how these tools perform and how they should be evaluated before being used in patient-facing settings,” said Horvat, who is also Associate Director of the Safar Center for Resuscitation Research at the University of Pittsburgh, named for Dr. Peter Safar, a pioneer of modern CPR.
The findings highlight an opportunity to study AI’s role in delivering guideline-based instructions—as a complement, not a replacement, for human responders. “This is about supporting people in high-stakes situations where human judgment is essential,” added Horvat. “The goal is to raise the floor of performance, not to replace trained professionals.”
Popular AI Models Show Promise But Fall Short
The research team first benchmarked popular AI models, including ChatGPT, Claude, Grok, Gemini, Llama, and Mixtral, on CPR coaching in simulated emergency scenarios, evaluating them against a checklist of criteria for delivering guideline-concordant CPR instruction in out-of-hospital cardiac arrest.
Across different situations, like drowning or collapsing while jogging, and across different patients, from toddlers to seniors, AI models performed well on the basics of CPR. On average, they scored 90% on essential steps, such as where to press on the chest and how fast to do it. Scores ranged from 79% (Gemini) to 97% (Grok and Claude). When it came to giving the best possible instructions to improve survival, performance dropped. On these more advanced steps, such as letting the chest fully rise between compressions, models averaged 70%. Scores ranged from 61% (Llama) to 75% (ChatGPT).
“In cardiac arrest, good is not good enough,” said Cameron Dezfulian, M.D, an adult and pediatric intensivist, senior faculty member at Baylor College of Medicine, and study coauthor. “Missing 10 to 30% of steps can be the difference between life and death.”
Building ChatCPR for Real Emergencies
Those gaps informed the development of ChatCPR, an open-source AI agent for CPR coaching grounded in 911 dispatcher training materials and CPR best practices. The system was iteratively engineered to address specific failure modes. In the same scenarios, ChatCPR scored 100% on simulated calls on both the basic steps and the more advanced steps needed to give someone the best chance for survival.
The critical question was whether it could work in real life. The team used a separate set of real, de-identified 911 calls that were publicly available. In these calls, dispatchers had already provided CPR instruction. The team then compared the dispatchers’ instructions to ChatCPR’s instructions.
ChatCPR Outperformed 911 Dispatchers in Every Comparison
“ChatCPR won every head-to-head comparison with human dispatchers,” explained Noor Majhail, B.S., an EMS responder and study coauthor. ChatCPR scored 15 percentage points higher than the dispatcher on basic CPR steps; specifically, dispatchers met 85% of the guideline steps, and ChatCPR met 100%. For more advanced steps, the gap was even larger; ChatCPR scored 99% while dispatchers scored 63% — a gap of 36 percentage points.
“ChatCPR excelled in patient assessment, chest compression depth and rate instructions, and recoil guidance areas where stressed, multitasking dispatchers most often faltered,” added Desai.
“This wasn’t about style. It was about strict adherence to CPR guidelines where precision matters most,” said Davey Smith, M.D., Professor at the UC San Diego School of Medicine, Director of the Altman Clinical and Translational Research Institute at UC San Diego and study coauthor. “ChatCPR addressed elements that dispatchers, under the stress and complexity of real calls, sometimes missed, presenting a valuable opportunity to translate AI into real-world healthcare practice.”
AI That Can Save Lives — With Safeguards
"No AI system is perfect, but the triangulation of problem identification, deep substantive expertise and AI is the crossroads for healthcare to unlock meaningful breakthroughs," said Rema Padman, Ph.D., Trustees Professor of Management Science and Healthcare Informatics in the Heinz College of Information Systems and Public Policy at Carnegie Mellon University and study coauthor.
The researchers say careful, real-world testing is still needed. They want to ensure the system is safe, works in chaotic settings, and is easy for people to follow. They also stress the importance of safeguards and the integration with existing 911 systems and continued human oversight.
Towards that goal, the team made ChatCPR open and free for anyone to use and study. They shared the complete system, how they tested it, and all related materials. “Any developer or organization can freely use, adapt, and deploy ChatCPR,” noted Mark Dredze, Ph.D., a Professor of Computer Science at Johns Hopkins and study coauthor. “We encourage researchers to refine, test, and improve this technology across all platforms to provide life-saving assistance for all.”
“AI could add value across the entire cardiac arrest response continuum, helping bystanders start CPR sooner, support dispatchers with standardized guidance, and assist clinicians and first responders with complex or scenario-specific instructions during training,” added Horvat.
Clear regulatory frameworks, as some of the author team has previously noted in an earlier JAMA piece, will also be essential as the tool moves from research to real-world use. “Today, bystanders have strong legal protections against civil and criminal prosecution when intervening to perform CPR, added Mike Hogarth, M.D., Professor of Medicine and Director of Informatics at the Altman Clinical and Translational Research Institute at UC San Diego and study coauthor. “How these protections extend to AI-enabled CPR is a challenge that needs to be addressed.”
“Ultimately, our work grounds AI hype in life-or-death reality,” concluded Ayers. “The real promise is closing the deadly gap between a person collapsing and lifesaving care beginning.”
The article, “An Artificial Intelligence–Enabled Cardiopulmonary Resuscitation Instructor” (doi:10.1001/jamainternmed.2026.1552) included Clifton Callaway, M.D., Ph.D. and Patrick M. Kochanek, M.D. (UPMC & University of Pittsburgh) as additional coauthors. It is available online on the JAMA Internal Medicine website.
“More than 350,000 Americans suffer out-of-hospital cardiac arrest each year, and survival sits at roughly 9%," says Nimit Desai, a research affiliate at the UC San Diego Qualcomm Institute, medical student at UC San Diego School of Medicine and co-author of a new study in JAMA Internal Medicine. "Given that only 2% of Americans are certified to perform CPR, when someone collapses, they call 911 and wait. ChatCPR could change that and begin to save lives.”
Credit
Courtesy of UC San Diego Herbert Wertheim School of Public Health and Human Longevity Science
ChatCPR offers AI-assisted CPR guidance, grounded in dispatcher protocols.
Credit
Courtesy of Qualcomm Institute UC San Diego
Video demonstration of ChatCPR [VIDEO]
This brief video, based on a real interaction, demonstrates an exchange between ChatCPR and a caller reacting to an emergency.
Credit
John W. Ayers, Qualcomm Institute, UC San Diego
Journal
JAMA Internal Medicine
Method of Research
Experimental study
Subject of Research
People
Article Title
An Artificial Intelligence–Enabled Cardiopulmonary Resuscitation Instructor
Article Publication Date
18-May-2026
COI Statement
Dr Desai reported receiving consulting fees from Pearl Health outside the submitted work. Dr Dredze reported receiving personal fees from Bloomberg LP, Good Analytics, and Medeloop outside the submitted work. Dr Hogarth reported receiving personal fees from LifeLink outside the submitted work. Dr Smith reported serving as a consultant for Model Medicines, Hyundai Biosciences, Capricor, Gilead, and Pfizer. Dr Callaway reported being a cofounder of Intellicardio, Inc outside the submitted work. Dr Ayers reported receiving personal fees and stock from Good Analytics, HealthWatcher, and Medeloop and building artificial intelligence analytical agents for health systems and life science companies outside the submitted work.
AI generates first complete models of proteins in motion
EPFL researchers have used a neural network to create all-atom models of proteins, as well as the dynamic movements that govern their function. Their comprehensive yet simplified approach resolves a major bottleneck in biology
Ecole Polytechnique Fédérale de Lausanne
video:
A simulated protein backbone (yellow) is gradually augmented with AI-generated snapshots from the LD-FPG, representing side chain positions and dynamic movements.
view moreCredit: LPCE LTS2 EPFL CC BY SA
Many drug and antibody discovery pathways focus on intricately folded cell membrane proteins. When molecules of a drug candidate bind to these proteins, like a key going into a lock, they trigger chemical cascades that alter cellular behavior. Understanding how proteins fold and move is therefore essential for developing drugs that interact well with their targets.
Artificial intelligence (AI) is a very useful tool to generate novel protein structures, but most systems – including Google DeepMind’s AlphaFold – focus on producing static ‘snapshots’ of proteins. Subtle rearrangements of atoms in structures called side chains, which influence a protein’s interactions with other molecules, are not captured.
Now, scientists in EPFL’s School of Life Sciences have teamed up with data processing experts in the School of Engineering to solve this problem. Researchers led by Patrick Barth of the Laboratory of Protein and Cell Engineering (LPCE) and Pierre Vandergheynst of the Signal Processing Laboratory (LTS2) have developed an AI-based generative framework called Latent Diffusion for Full Protein Generation (LD-FPG), which produces complete, all-atom structural ensembles of proteins and their movements.
“Proteins are like tiny machines that dance and switch on and off to work but generating this ‘movie’ in full detail has been an unsolved challenge,” says LPCE researcher Aditya Sengar. “Our LD-FPG framework is the first to do this. Instead of trying to predict the exact coordinates of atoms in space, our model learns a low-dimensional map of the protein's shape changes. This conceptual shift is what makes generating all-atom dynamics possible.”
The new framework can notably generate the full range of motion for complex drug targets like G-protein coupled receptors (GPCRs): a focus of the global drug development industry.
“LD-FPG opens the door to designing new medicines that target a protein's dynamic behavior, not just its shape. Our work represents a new paradigm for computational biology, and a meaningful step forward at the interface of AI and structural biology,” says Barth. The work has been published in the Proceedings of NeurIPS 2025.
Capturing a protein’s dance
Because systems like AlphaFold use AI to predict the spatial position of every atom in a protein, they require vast amounts of computing power and biology and computer science expertise. LD-FPG simplifies this problem using something called a graph neural network (GNN). The GNN treats each protein like a mathematical graph, where atoms represent ‘nodes’ and the bonds between them represent ‘edges’. Using this low-level representation, it essentially compresses protein structure data into a simplified, or latent, map.
Next, an AI model studies this map and ‘learns’ the representations of the protein’s structure and movements. Once trained, the model generates latent data for entirely new structures. Finally, these simplified data are converted back into high-resolution proteins – complete with side chains and dynamic movements.
In one experiment, the team generated high-fidelity, dynamic representations of the dopamine D2 receptor in both its active and inactive states. This protein detects the neurotransmitter dopamine and controls key cellular responses, making it one of the most-studied GPCRs. The researchers have published this dataset with open access to facilitate further research.
“In addition to enhancing biological understanding, we believe our work will help improve virtual screening processes for proteins, which currently involve a lot of trial and error, thereby accelerating drug discovery,” Sengar says.
Going forward, the team aims to streamline the AI framework for even greater accuracy and realism, and to enable it to model larger proteins. But Vandergheynst emphasizes that high-quality data will remain the bedrock of success: “Many assume that feeding massive datasets to AI models will automatically solve scientific problems or replace researchers. However, much of that data is noisy or poorly evaluated. We need human scientists to produce the clean data and rigorous benchmarks AI requires, much like we need journalists to safeguard against disinformation."
Article Title
Generative Modeling of Full-Atom Protein Conformations using Latent Diffusion on Graph Embeddings
A simulated protein backbone (yellow) augmented with AI-generated snapshots. LPCS LTS2 EPFL CC BY SA
Credit
LPCS LTS2 EPFL CC BY SA
Penn researchers create AI tool to speed antibiotic discovery
By combining generative AI with Bayesian optimization, ApexGO suggests molecular edits that can turn weak antibiotic candidates into more potent ones.
image:
A 3D-printed example of the kind of antibiotic peptide the researchers generated using AI, held in a server room at the University of Pennsylvania.
view moreCredit: Sylvia Zhang, Penn Engineering
Researchers at the University of Pennsylvania have developed ApexGO, a novel, AI-powered method for turning promising but imperfect antibiotic candidates into more potent ones.
Unlike many existing AI approaches to antibiotic discovery, which screen large databases for molecules that might work, ApexGO starts with a small number of imperfect candidates and improves them step by step, using a predictive algorithm to evaluate each modification and guide the next.
“Antibiotic discovery is fundamentally a search problem across an enormous molecular space. ApexGO gives us a way to navigate that space with far more direction,” says César de la Fuente, Presidential Associate Professor in Bioengineering and in Chemical and Biomolecular Engineering in the School of Engineering and Applied Science, in Psychiatry and Microbiology in the Perelman School of Medicine and in Chemistry in the School of Arts & Sciences, and co-senior author of a new paper describing the method in Nature Machine Intelligence.
“ApexGO begins with a promising but imperfect peptide,” explains de la Fuente, referring to a short string of amino acids, “proposes precise edits, predicts whether those changes are likely to enhance antimicrobial activity, and then keeps moving toward versions that are more likely to work when we make and test them.”
Laboratory tests against disease-causing bacteria supported ApexGO’s predictions: 85% of the AI-generated molecules halted bacterial growth, while 72% outperformed the peptides from which they were derived. In mice, two antimicrobial peptides created by ApexGO reduced bacterial counts at levels comparable to polymyxin B, an FDA-approved antibiotic used as a last-resort treatment for some drug-resistant infections.
“What is striking is that ApexGO’s predictions held up in the real world,” says Jacob R. Gardner, Assistant Professor in Computer and Information Science (CIS) and the paper’s other senior co-author. “ApexGO was optimizing against another computer model, so one concern was that it might find molecules that looked good to the model but failed in the lab. Instead, the majority of the molecules it designed actually worked.”
From Screening Molecules to Making New Ones
For years, the de la Fuente lab has looked for antibiotic candidates in unlikely places, from frog secretions to ancient microbes. Two years ago, the group released APEX, an AI model that predicts whether or not a given peptide is likely to have antimicrobial properties.
“APEX helped us find promising antibiotic candidates in enormous biological datasets,” says Marcelo Torres, Research Assistant Professor of Psychiatry in the Perelman School of Medicine and co-first author of the paper, referring to work that revealed antibiotic candidates everywhere from woolly mammoths to giant sloths. “ApexGO takes the next step: once we have a promising molecule, it helps us ask how to make it better.”
That’s where Gardner’s lab comes in. The group specializes in methods like Bayesian optimization, which helps AI systems explore large numbers of possible solutions efficiently. “It would be impossible to test every possible peptide,” says Yimeng Zeng, a doctoral student in CIS and co-first author of the paper. “Bayesian optimization helps the model make informed choices about what to try next, balancing candidates that look promising with candidates that could teach the model something new.”
Essentially, one part of ApexGO — short for APEX Generative Optimization — suggests molecular tweaks, while the previously published APEX model predicts whether those changes are likely to increase antimicrobial activity. ApexGO then uses those predictions to guide the next round of proposed edits. “If a region of the search space looks promising, the model can spend more effort exploring nearby variants,” says Zeng. “But it can also move into less certain regions, where there may still be hidden improvements.”
Searching More Systematically
Until now, the researchers point out, antibiotics have largely been found by accident. The most famous example is also the first: penicillin, which Alexander Fleming discovered after noticing that mold in a petri dish was restricting the growth of bacteria. “In a sense, we’ve been incredibly lucky,” says de la Fuente. “ApexGO points to a more systematic way forward.”
The space of all possible antimicrobial peptides is huge: Like searching a vast forest for something small or rare, finding an antibiotic peptide is normally prohibitively time-consuming. Even a short peptide can be modified in an enormous number of ways, making it impossible for researchers to synthesize and test every possible version by hand.
That ApexGO could identify antibiotic candidates with laboratory activity against disease-causing bacteria, simply by searching this space computationally, points to a different approach. “We ran ApexGO for a few months and found hundreds of candidates,” notes Gardner. “If we ran that process for a year, how many thousands of these could we find?”
“This result points toward a future in which we can optimize molecules for a desired function in a fraction of the time,” adds de la Fuente, “using machines to guide discovery through chemical spaces too vast for humans to explore by trial and error.”
Future Directions
While some of the molecules proposed by ApexGO showed promising antibiotic activity, the researchers emphasize that even the best-performing peptides are still early-stage candidates. Before any could be used to treat infections in humans, they would need to be further optimized for safety, stability and how long they remain active in the body.
Still, the study suggests that AI can help researchers decide which molecules are worth making and testing in the first place. Instead of synthesizing one candidate after another by trial and error, tools like ApexGO could help narrow the search to molecules more likely to work.
For de la Fuente, that approach could eventually extend beyond antibiotics. “In this case, we wanted to optimize peptides for antimicrobial activity,” he says. “But you could imagine applying the same idea to peptides with other biological functions, like modulating the immune system or targeting tumors.” Gardner’s lab is already exploring related approaches using AI agents, which may be able to draw on scientific knowledge and reason through design choices.
“The larger idea is that AI can help scientists search spaces that are too large to explore by hand,” says Gardner. “ApexGO is one example of that. The next generation of tools may be able to explore these spaces in even more flexible ways.”
“ApexGO shows that AI can do more than predict which molecules might work: it can help us improve them,” adds de la Fuente. “At a time when antibiotic resistance is rising worldwide, we need technologies that help us move faster from an idea to a real therapeutic candidate. ApexGO is an important step toward that future.”
This study was conducted at the University of Pennsylvania and supported by the National Institutes of Health (R35GM138201), the Defense Threat Reduction Agency (HDTRA1-21-1-0014), the National Science Foundation (IIS-2145644, DBI-2400135) and a National Science Foundation graduate research fellowship.
Additional co-authors include co-first author Fangping Wang of the Perelman School of Medicine, School of Engineering and Applied Science, and School of Arts & Sciences; and Natalie Maus of the School of Engineering and Applied Science.
The researchers ran their new AI tool ApexGO for months on Penn servers like this one, and the model produced hundreds of new antibiotic candidates.
Credit
Sylvia Zhang, Penn Engineering
Journal
Nature Machine Intelligence
Method of Research
Experimental study
Subject of Research
Animals
Article Title
A generative artificial intelligence approach for peptide antibiotic optimization
Article Publication Date
13-May-2026
COI Statement
César de la Fuente-Nunez is a co-founder and scientific advisor to Peptaris, Inc., provides consulting services to Invaio Sciences and is a member of the Scientific Advisory Boards of Nowture S.L. and Phare Bio. The de la Fuente Lab has received research funding or in-kind donations from United Therapeutics, Strata Manufacturing PJSC, and Procter & Gamble, none of which were used in support of this work. Jacob Gardner serves on the scientific advisory board of BigHat Biosciences, Inc. Marcelo D. T. Torres is a co-founder and scientific advisor to Peptaris, Inc. All the other authors declare no competing interests.
