Hacker la bombe ? Ce que l’IA Mythos révèle du pari de la dissuasion nucléaire

Source: The Conversation – France in French (3) – By Thomas Fraise, Postdoctoral research fellow, University of Copenhagen; Sciences Po

Claude Mythos est une intelligence artificielle de pointe, développée par la société Anthropic, dont les capacités avancées en cybersécurité offensive suscitent autant de fascination que d’inquiétude. Gguy/Shutterstock

L’apparition de l’IA Mythos, présentée comme capable de détecter rapidement des failles informatiques majeures, montre que les capacités offensives dans le cyberespace progressent très vite. Cette évolution pourrait rendre les systèmes nucléaires plus vulnérables et augmenter le risque d’erreurs, de sabotage ou d’escalade accidentelle.


En 1983, le film WarGames imaginait un adolescent qui, entré par accident dans un système informatique du Pentagone, déclenchait un programme de simulation, lequel était interprété comme un prélude à une guerre nucléaire. Le film avait tant marqué Ronald Reagan qu’il avait interrogé ses conseillers sur la possibilité d’une telle intrusion dans les systèmes américains les plus sensibles. Une semaine plus tard, la réponse était venue : « Monsieur le Président, le problème est bien plus grave que vous ne le pensez. »

Les politiques autour de l’armement nucléaire reposent sur une série de paris, souvent lointains, sur l’avenir de la dissuasion nucléaire. D’abord, les pays dotés de l’arme nucléaire considèrent que la peur de leur riposte suffira toujours à empêcher un adversaire de les attaquer en premier, et qu’ils disposeront toujours de l’expertise et de la chance nécessaires pour prévenir des explosions accidentelles. Ils estiment donc que la possession de l’arme nucléaire sera pour eux, au cours des décennies à venir, source de sécurité et non d’insécurité.

Or, comme nous le montrons avec mes collègues Sterre van Buuren et Benoît Pelopidas, il existe plusieurs scénarios futurs plausibles dans lesquels posséder des armes nucléaires engendrera plus de coûts réels que de bénéfices potentiels dans un monde où la température aura augmenté de plusieurs degrés. Maintenir un arsenal crédible et sûr exigera de faire des choix budgétaires, au détriment d’autres dépenses rendues urgentes par la crise climatique. L’univers des risques existentiels pour un Etat, qui pourrait justifier l’usage de l’arme nucléaire, peut aussi s’élargir. Des experts s’inquiètent que des risques de pénurie d’eau au Pakistan et en Inde ne deviennent un terrain fertile pour un conflit menant à une escalade nucléaire.L’univers des risques existentiels pour un État, qui pourrait justifier l’usage de l’arme nucléaire, peut aussi s’élargir. Des experts s’inquiètent que des risques de pénurie d’eau au Pakistan et en Inde ne deviennent un terrain fertile pour un conflit menant à une escalade nucléaire.

Mais il existe un autre pari, bien plus implicite : celui que les arsenaux nucléaires, qui sont des systèmes technologiques complexes et hautement digitalisées, ne possèdent aucune vulnérabilité cyber qui pourrait être exploitée par un acteur souhaitant empêcher son fonctionnement normal.

La récente percée de l’intelligence artificielle Mythos révèle à quel point les conditions de ce pari peuvent changer sur le long terme.

Mythos et l’avenir de la cybersécurité

Le 7 avril 2026, la compagnie Anthropic – qui commercialise la série de grands modèles de langages (LLM) Claude – annonçait la création de son nouveau modèle d’intelligence artificielle (IA) : « Mythos ». Ce modèle, qui n’a pas été mis sur le marché mais rendu disponible à un groupe de travail restreint composé d’une douzaine des principaux géants américains de la technologie (Google, Microsoft, Apple, NVidia, Amazon Web Services, etc.), obtiendrait un taux de succès sans précédent en matière de détection des failles dans les systèmes informatiques.

Mythos aurait ainsi été capable de détecter, avec un impressionnant taux de succès, des failles « zero-day » dans différents navigateurs informatiques, logiciels ou systèmes d’exploitation. Une faille « zero-day » est une faille de sécurité critique dans un système d’information, contre laquelle aucune protection n’existe pour l’instant, rendant ainsi possible une attaque laissant un délai de « zéro jour » pour réagir. Selon Anthropic, Mythos aurait réussi à développer en un temps record (sans doute moins d’une journée) des méthodes permettant d’exploiter ces failles avec un taux de succès de 72,4 %, largement supérieur aux autres modèles existants.

Si ces informations proviennent de la compagnie elle-même – qui a tout intérêt à exagérer les résultats –, certaines preuves publiques ont toutefois été apportées. Sylvestre Ledru, le directeur de l’ingénierie chez Mozilla responsable du navigateur Firefox, a déclaré que Mythos avait permis de découvrir un nombre « proprement hallucinant » de vulnérabilités dans leurs logiciels. Une faille de sécurité vieille de près de vingt-sept ans, ayant survécu à un grand nombre d’audits, a par exemple été découverte dans un système d’exploitation libre très utilisé par des services de sécurité informatique, OpenDSB.

Mythos révèle un problème de fond : l’augmentation des capacités offensives – non seulement des États, mais aussi d’acteurs privés comme des cybercriminels – dans le cyberspace risque d’être accélérée par le développement de l’IA, et une incertitude émerge quant à la capacité des acteurs défensifs à réagir suffisamment vite pour corriger les vulnérabilités existantes.

Même dans le cas où Mythos ne serait pas à la hauteur des performances annoncées, le développement des LLM depuis le début des années 2020 a montré à quel point leurs performances s’améliorent vite. Nous faisons donc face à une accélération du développement des capacités offensives et de la diffusion de celles-ci à un nombre d’acteurs plus large. Cela signifie une potentielle tendance à la hausse de la probabilité de succès d’une cyberattaque, ainsi qu’une augmentation du nombre absolu de ces attaques.

La vulnérabilité des arsenaux nucléaires

Pour comprendre la vulnérabilité des armes nucléaires aux cyberattaques, il faut avoir à l’esprit le fait que par « arsenal nucléaire », on entend bien plus qu’un stock de têtes nucléaires. Le fonctionnement normal des arsenaux nucléaires modernes repose sur une large configuration de technologies : têtes nucléaires, missiles permettant de transporter ces armes, technologies de communication (permettant d’assurer que l’ordre soit transmis depuis le président jusqu’à l’opérateur chargé d’actionner ces armes), ainsi qu’un ensemble de technologies d’alerte avancée servant à surveiller le ciel à la recherche de potentiels signaux d’une attaque nucléaire adverse. Ces éléments doivent être capables de communiquer entre eux pour assurer le contrôle de ces armes.

Et ils sont plus nombreux qu’on pourrait le penser. Comme le note Herbert Lin, chercheur à l’Université de Stanford et auteur d’une étude sur les cybermenaces contre les armes nucléaires, la métaphore du « bouton nucléaire » est simplifiée : une fois que le président appuie dessus, un ensemble de « cyber-boutons » doivent être pressés pour déclencher les opérations nucléaires, et les contrôler – autant d’échelons où des attaques informatiques pourraient s’insérer pour empêcher, par exemple, l’arrivée de l’information pertinente.

Le président pourrait ne pas recevoir suffisamment d’informations – ou ne pas en recevoir du tout – pour déterminer qu’une attaque est en cours. Ou alors, il pourrait ne pas être en mesure de communiquer à ses forces sous-marines l’ordre de tir. Pis, on pourrait voir se réaliser le scénario catastrophe imaginé depuis les années 1950 : un faux ordre de tir pourrait être communiqué aux opérateurs de missiles.

Les scénarios n’ont pas besoin d’être aussi radicaux : l’ordre pourrait être communiqué, mais avec un retard, ou ne pas être communiqué à l’ensemble des forces, menant à une riposte moindre que celle désirée. Ladite riposte pourrait être bloquée : en 2010, un centre de commandement américain a perdu la communication avec une cinquantaine de missiles nucléaires pendant près d’une heure. Un adversaire pourrait savoir tirer profit de telles failles.

Alternativement, une cyberattaque d’ampleur menée par des acteurs non étatiques pourrait créer l’impression qu’un adversaire cherche à s’en prendre à notre arsenal nucléaire, créant un risque d’escalade par « inadvertance ».

On peut aussi imaginer des actions cyber contre les armes elles-mêmes, le hardware plutôt que le software de l’arsenal. Bien sûr, les acteurs de la sécurité nucléaire ne se contentent pas d’attendre qu’une attaque survienne sur l’un de ces systèmes. Ils développent et testent leurs capacités défensives de manière continue. Le problème est que la complexité des systèmes existants ne permet pas d’affirmer avec certitude qu’il n’existe « aucune vulnérabilité ».

C’est James Gosler, ancien responsable de la sécurité informatique des systèmes nucléaires américains au sein du laboratoire Sandia, qui l’affirme : à partir des années 1980, du fait de la complexification exponentielle des composants internes aux armes nucléaires, « vous ne pouvez désormais plus affirmer que l’ensemble des microcontrôleurs (destinés à assurer le fonctionnement du mécanisme déclenchant l’explosion) sont invulnérables ».

Cela ne signifie pas non plus que des vulnérabilités existent nécessairement. Mais cela veut dire qu’aucun acteur n’est en mesure de savoir s’il y en a. Alors, faut-il craindre que l’arsenal nucléaire français, ou bien n’importe quel autre arsenal ciblant la France, soit « hacké » dans le futur ?

En fait, on ne sait pas. Des scénarios de ce type sont de l’ordre du possible : il n’existe pas de large système d’information complexe dont on puisse garantir, avec une totale certitude, la fiabilité totale. L’évolution des outils permettant des cyberattaques, et leur potentielle diffusion auprès d’un large nombre d’acteurs étatiques et non étatiques, rend ce type de scénario futur potentiellement plus probable et, dans tous les cas, plausible.

Un nouveau pari sur le futur

Mythos met en lumière une nouvelle modalité du pari nucléaire, née du développement des nouvelles technologies et de leur intégration aux arsenaux nucléaires.

Nous parions d’abord sur l’absence de vulnérabilité au sein de ces systèmes – alors même qu’il est impossible de mesurer cette probabilité avec certitude. Elle évolue avec le temps, au rythme des systèmes mis à jour, remplacés, connectés à d’autres. Si une vulnérabilité existe malgré tout, nous parions ensuite sur le fait que, en temps voulu, l’évolution des capacités offensives dans le cyberespace sera constamment égalée, et toujours à temps, par l’évolution des capacités défensives – y compris à l’ère de l’intelligence artificielle. Là encore, cette probabilité est elle aussi indéterminable, puisque le développement de capacités défensives est réactif : il se fait en fonction de la connaissance que l’on a de la nature des capacités offensives et des vulnérabilités existantes, qui sont indéterminables. On fait donc le pari que nos défenses, et celle des autres États dotés d’armes nucléaires, suffiront.

Nous faisons donc le pari que nos défenses contre les cyberattaques, et celle des autres États dotés d’armes nucléaires, seront suffisantes. Dans le cas contraire, alors on fait le pari que la chance sera de notre côté et que les vulnérabilités existantes ne seront pas détectées – comme celle qui existait depuis 27 ans dans le code d’OpenDSB. Il s’agit d’un pari sur la chance puisque dans ce scénario, c’est l’incapacité ou l’absence de volonté adverse, sur laquelle nous n’avons aucun contrôle, à développer des capacités efficaces qui nous sauve.

La capacité des pratiques existantes de contrôle à remplir leur tâche est rendue plus incertaine par l’arrivée de grands modèles d’IA capables de détecter des vulnérabilités et concevoir des cyberattaques de façon massive et automatisée. Faire le choix d’une politique de sécurité fondée sur les armes nucléaires revient à parier sur le fait que, dans le futur comme par le passé, la chance restera toujours de notre côté.

The Conversation

Ce travail a été financé par le Conseil Européen de la Recherche (ERC) au titre du programme-cadre de l’Union européenne pour la recherche et l’innovation Horizon Recherche (projet RITUAL DETERRENCE, convention de subvention n°101043468).

ref. Hacker la bombe ? Ce que l’IA Mythos révèle du pari de la dissuasion nucléaire – https://theconversation.com/hacker-la-bombe-ce-que-lia-mythos-revele-du-pari-de-la-dissuasion-nucleaire-281557

Biais de la recherche médicale : l’Afrique absente des essais cliniques

Source: The Conversation – in French – By Bamba Gaye, MD, MPH, MSc, PhD, Executive Director, Alliance for Medical Research in Africa (AMedRA) | Adjunct Professor at Biomedical Informatics, Emory School of Medicine in Atlanta, Georgia, USA |, Université Cheikh Anta Diop de Dakar

La médecine moderne se targue d’être une science universelle, fondée sur les données issues des essais cliniques.

Mais il existe un biais dans la recherche médicale. Alors que l’Afrique représente environ 25 % de la charge de morbidité du monde et 19 % de la population mondiale. Pourtant, sa population est largement absente des essais cliniques.

L’ampleur de cette exclusion est révélée par une étude phare portant sur 2 472 essais contrôlés randomisés publiés dans le monde entier entre 2019 et 2024.

J’ai dirigé cette équipe de chercheurs, qui a passé au crible les publications médicales les plus influentes dans le monde monde afin de quantifier la représentation africaine. Parmi celles-ci figuraient le New England Journal of Medicine, The Lancet, le Journal of the American Medical Association, Nature Medicine et le British Medical Journal. L’étude portait également sur trois revues de référence en cardiologie : Circulation, l’European Heart Journal et le Journal of the American College of Cardiology.

Je suis médecin-chercheur et je travaille à la croisée de l’épidémiologie cardiométabolique et de la science des données biomédicales. Je me consacre également aux études de population à grande échelle en Afrique et à la prévention cardiovasculaire fondée sur les données.

Les essais cliniques randomisés constituent la pierre angulaire de la médecine factuelle. Introduits au milieu du XXe siècle, ils évaluent rigoureusement la sécurité et l’efficacité des traitements en répartissant aléatoirement les participants dans différents groupes. Cette méthode vise à minimiser les biais. De tels essais ont joué un rôle central dans les avancées médicales majeures, des thérapies cardiovasculaires aux vaccins. Ils continuent de guider les décisions cliniques et le développement de nouveaux traitements à travers le monde.




Read more:
Accords sanitaires Afrique–États-Unis : les alertes d’un virologue


Ce que nous avons découvert

Nos résultats révèlent un profond déséquilibre dans le paysage mondial de la recherche clinique. Parmi les cinq revues médicales les plus prestigieuses dans le monde, seuls 3,9 % des essais ont été menés exclusivement en Afrique. En matière de santé cardiovasculaire, ces chiffres tombent à un niveau statistiquement insignifiant. Parmi les essais majeurs publiés dans les principales revues de cardiologie, seules deux études (0,6 %) ont été menées uniquement sur le sol africain.

C’est un problème de fiabilité scientifique. Lorsque les essais cliniques excluent les populations africaines, ils produisent des données qui manquent de « validité externe ». Cela fait référence à la capacité des résultats d’une étude à être généralisés au-delà des participants. La question est de savoir si les conclusions d’un essai clinique resteront valables lorsqu’elles seront appliquées à des populations, des contextes ou des conditions réelles différents.

Sans cette validité, les médecins appliquent des résultats incertains à des millions de patients chaque jour.

La médecine moderne ne peut prétendre être universelle si des populations entières restent invisibles dans la base de données probantes. La biologie, les systèmes de santé et les schémas épidémiologiques ne sont pas identiques partout dans le monde.




Read more:
Protocole d’accord avec les États-Unis : le Sénégal teste un nouveau modèle de financement de la santé


Le fossé et son importance

De nombreux traitements utilisés sur le continent reposent sur des données générées auprès de populations non africaines, ce qui soulève des inquiétudes quant à leur applicabilité.

De plus, la plupart des essais menés en Afrique se concentrent encore sur les maladies infectieuses, malgré le fardeau croissant des maladies non transmissibles telles que les maladies cardiovasculaires.

De nouvelles données montrent que la génétique, l’environnement et l’alimentation peuvent modifier radicalement la façon dont un organisme réagit à un médicament. Il n’est donc pas logique, d’un point de vue médical, qu’un continent entier soit exclu du réseau des essais cliniques.

Il existe également des preuves montrant que certains traitements présentent des profils de sécurité différents chez les patients noirs. Le diabète et la goutte n’en sont que deux exemples. Il en va de même pour certains médicaments courants contre l’hypertension, tels que les inhibiteurs de l’enzyme de conversion de l’angiotensine (ECA). Des recherches montrent qu’ils présentent un risque trois à quatre fois plus élevé d’effets secondaires graves et potentiellement mortels chez les personnes d’origine africaine par rapport aux autres populations.

Lorsque les essais cliniques excluent certaines populations, les médecins sont contraints d’extrapoler les résultats obtenus sur une population et de les appliquer à une autre.

L’étude met également en évidence un décalage dangereux entre le financement mondial de la recherche et l’évolution de la réalité sanitaire en Afrique. Les nouvelles données montrent que près de 76 % des essais menés exclusivement en Afrique portaient sur des maladies infectieuses. Or, le continent connaît actuellement une transformation épidémiologique majeure. Les maladies non transmissibles – maladies cardiaques, accidents vasculaires cérébraux et diabète – représentent désormais environ 38 % de tous les décès dans de nombreux pays africains.

La classe moyenne en Afrique a triplé pour atteindre 300 millions de personnes, contre environ 100 millions au début des années 2000. De plus en plus de personnes vivent désormais suffisamment longtemps avec des modes de vie qui augmentent le risque de maladies chroniques telles que les maladies cardiaques, le diabète et l’hypertension. Par conséquent, il existe un besoin et un marché croissants pour des traitements à long terme permettant de prendre en charge ces maladies, plutôt que pour des thérapies à court terme contre les infections. Pourtant, les essais cardiovasculaires continuent d’être découragés.

Même au sein du continent, les données révèlent de véritables zones d’ombre en matière d’information. L’Afrique du Sud représente plus de 62 % de tous les essais menés sur le continent. L’Afrique centrale, une région qui compte plus de 180 millions d’habitants, était pratiquement absente des registres mondiaux de la recherche. Elle a contribué à moins de 3 % de la production limitée d’essais cliniques du continent.

Parmi les raisons possibles, on peut citer les décennies d’investissements cumulés de l’Afrique du Sud, qui se traduisent par des pôles universitaires plus solides, une meilleure gouvernance de la recherche, des unités d’essais expérimentées et des relations plus établies avec les promoteurs. D’autres régions sont confrontées à des obstacles tels que des institutions de recherche disposant de moins de ressources, un accès plus limité aux plateformes d’essais cliniques, et parfois des problèmes linguistiques et de publication qui peuvent réduire leur visibilité dans les revues de premier plan.

Cette inégalité s’étend à la hiérarchie scientifique elle-même. Même lorsque des sites africains sont inclus dans de grands essais cliniques multiconinentaux, ils sont souvent relégués au rôle de « centres de recrutement » plutôt qu’à celui de partenaires scientifiques.

Notre étude a révélé que les scientifiques africains ne dirigeaient que 3,6 % des essais sur plusieurs continents incluant un site africain.




Read more:
L’Afrique perd ses professionnels de santé au pire moment : les racines coloniales d’une crise


Vers une nouvelle ère pour la science africaine

L’Afrique ne devrait pas se contenter d’être un simple lieu où l’on mène des études.

Elle doit être un lieu où la recherche est conçue, dirigée et interprétée. Le modèle actuel crée un cycle de dépendance vis-à-vis de l’extérieur, dans lequel les institutions internationales gèrent le financement et les données. Cela rend les systèmes de recherche locaux fragiles et incapables de traduire les données probantes en politiques nationales.

Il est nécessaire de mettre en place un financement « réservé » pour la recherche dirigée par des Africains, de développer des réseaux régionaux d’essais cliniques et d’imposer aux revues médicales de rendre compte de la diversité des populations participant aux essais.

Des signes indiquent une dynamique croissante. Des organisations telles que l’Alliance pour la recherche médicale en Afrique s’efforcent de former une nouvelle génération de chercheurs africains. L’Afrique doit créer un écosystème de recherche trop important pour que la communauté internationale puisse l’ignorer.

The Conversation

Bamba Gaye, MD, MPH, MSc, PhD does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

ref. Biais de la recherche médicale : l’Afrique absente des essais cliniques – https://theconversation.com/biais-de-la-recherche-medicale-lafrique-absente-des-essais-cliniques-282184

The 2025 Sir Paul Curran award for academic journalism goes to Jeremy Howick

Source: The Conversation – UK – By Michael Parker, Director of Operations, The Conversation

Jeremy Howick, Professor of Empathic Healthcare at the University of Leicester, receives the Sir Paul Curran award for 2025 from Lady Helen Curran, at the event at Bayes Business School. The Conversation

Jeremy Howick, Professor of Empathic Healthcare and Director of the Stoneygate Centre at the University of Leicester has been named this year’s winner of the Professor Sir Paul Curran Award for Excellence in Academic Journalism.

The prize is awarded annually to an academic who has shown exceptional skill, dedication and engagement in communicating their knowledge to readers through their contributions for The Conversation.

Jeremy has written 26 articles that have garnered 775,000 pageviews since the first piece in 2016, including translations into French and Portuguese. His articles have looked at placebos, the effects of empathy and empathic treatment by clinicians, and medical safety, among others.

Presented this year by Lady Helen Curran in Sir Paul’s absence, we were delighted to welcome around 80 authors who had written for The Conversation in 2025 to join us at Bayes Business School, City St George’s, University of London, for drinks and the opportunity to meet the editors they work with and colleagues from across the sector.

Jeremy said: “It is a great honour to win this prestigious award. Thank you to The Conversation for the wonderful work they do, to my editor Clint Witchalls and to Stephen Khan for your help over the years.

“I have written for The Conversation for 10 years, and it has been an important part of my life’s work. I have always viewed public communication as a duty. My research as an academic has been funded by the MRC, NIHR, and other public bodies, meaning that the taxpayer has often paid my bills. The Conversation gives me a forum through which I can explain my research to the public.

“But also I’ve always been motivated to change practice. You can do that by becoming a politician or policymaker, but I’m not patient enough for that. The other way is to inform the public to create a groundswell for change. At the Stoneygate Centre for Empathic Healthcare in Leicester, and before that at the Oxford Empathy Programme, I’ve attempted to do just that by working to ensure that all healthcare consultations include a dose of empathy. The evidence is starting to show that we are succeeding at making a real difference, and The Conversation has contributed to that real-life impact. So thank you once again for this amazing award. I’m humbled and honoured.”

Senior Health Editor Clint Witchalls said: “I remember Jeremy’s first piece: Why doing good can do you good, about how doing good things for others doesn’t just feel nice but can actually make you healthier. Studies showed that volunteers have lower stress levels, healthier hearts, and even a brain that rewards them with feel-good chemicals for being kind. At the time Jeremy was at the University of Oxford and focused on researching the placebo and nocebo effects, which he has also written about for us. I can highly recommend Jeremy’s book: The Power of Placebos.”

Commendations

Highly Commended for his work was archaeology researcher Stephan Blum, from the University of Tübingen in Germany. Perhaps understandably he was not able to attend on the day, but his certificate is winging its way eastward.

Senior Arts Editor Anna Walker said: “Not only is Stephan a delight to work with, but he instinctively understands what The Conversation needs. His ability to connect the ancient world to contemporary concerns is a defining strength of his writing. Whether exploring environmental pressures behind the fall of Troy or reassessing long-held assumptions about early trade and wine production, he shows how the distant past can illuminate present-day issues, presenting history not as something static or remote but as a dynamic field that continues to shape how we understand the world today.”

Responding to the award, Stephan wrote: “Writing for The Conversation has been a tremendously rewarding experience, and I truly appreciate the opportunity to contribute. It means a great deal to see the work recognised in this way.”

Commended for their work on the thorny topic of climate finance were Meilan Yan of Loughborough University and Narmin Nahidi of the University of Exeter (who was also not able to attend).

Meilan Yan, Senior Lecturer in Financial Economics at Loughborough University, receives her commended certificate from Lady Helen Curran at the 2025 Sir Paul Curran awards.
The Conversation

Senior environment editor Anna Turns said: “I first worked with Meilan in May 2025 when compiling a user-friendly guide to climate finance – a notoriously complex and abstract topic. Climate risk is one of Meilan’s biggest worries and this is very much something that gets overlooked among a chaotic landscape of geopolitical unrest. Her pieces bridge the gap between climate science and financial risk in tangible and relatable ways.

“Narmin Nahidi, who we also discovered through working on the glossary of climate finance, explained many terms for us and has since tuned in to pitching timely and topical stories, is always been keen to collaborate on edits and a pleasure to work with.

Editor-in-Chief Stephen Khan thanked Lady Curran: “Professor Sir Paul’s ongoing commitment to The Conversation is greatly appreciated and we were delighted Lady Curran was able to meet the authors shortlisted this year and make the presentations.

“We’ve had thousands of fantastic articles and podcast contributions from across the academy over the past 12 months, taking expert knowledge to millions of people from all walks of life around the world.

“These authors authors showed a particular ability to make complex research accessible, engaging and genuinely useful. Jeremy Howick’s writing exemplifies the very best of The Conversation’s mission: rigorous evidence communicated with clarity, warmth and public purpose.”

As ever, a huge thank you to Jeremy, Stephan, Meilan and Narmin for their work with The Conversation over the years, and to all our authors – without whose efforts there would be no conversation.

The Conversation

ref. The 2025 Sir Paul Curran award for academic journalism goes to Jeremy Howick – https://theconversation.com/the-2025-sir-paul-curran-award-for-academic-journalism-goes-to-jeremy-howick-282413

Donald Trump’s chaotic mess: When U.S. power serves the ‘sultan,’ global rules erode

Source: The Conversation – Canada – By Christopher Collins, Fellow, Geopolitics, Cascade Institute, Royal Roads University

Historically, the United States hasn’t always been easy to deal with, but it was consistent. Even countries that disagreed with American policies knew there was a logic underlying its actions, and this predictability gave the country some credibility.

But now, under U.S. President Donald Trump’s second administration, American foreign policy has become haphazard and contradictory, driven by a leader who believes his ability to exercise power around the world is constrained only by his own morality.

This is new and, for observers around the world, perplexing. As Canadian Prime Minister Mark Carney recently said: “Washington has changed. There is almost nothing normal now in the United States.”

Trump maelstrom

Some, like U.S. Vice President JD Vance, are labouring to erect a retroactive, pseudo-intellectual scaffolding around this chaotic mess, seeking to frame it as a coherent doctrine. But it’s become increasingly clear there’s no grand plan, just a Trumpian maelstrom of impulsive reactions, extractive transactions and personal grudges that shift with the news cycle.

To understand this political dysfunction, a German thinker from more than 100 years ago, Max Weber, offers a helpful guide.

Most famous today for his theory of “the Protestant work ethic,” Weber’s writing also explored the concept of “patrimonialism.”

This is a system of governance in which a ruler treats the state as personal property, governs by whim and uses the state’s resources to reward cronies and enrich family. Drawing largely on his understanding of the Ottoman Empire, Weber called the most extreme form of this system “sultanism.”

Reading Weber today, it seems the best description of how the U.S. engages the wider world could be termed “sultanism with American characteristics.”

Loyalty over experience

Consider Iran. Following the start of Operation Epic Fury, the Trump administration cycled through so many conflicting war aims that CNN was able to assemble a montage of the contradictions.

Senior administration officials worked feverishly to build a strategy around the operation, but it soon became clear that this “war of choice” was started based on little more than the president’s whim.




Read more:
Vietnam ruined Lyndon B. Johnson’s political career. Will Donald Trump face the same fate over Iran?


Weber’s framework extends to the people around Trump. In sultanistic systems, staff are selected based on loyalty, not merit, and serve the ruler, not the state.

As Weber wrote, this leads to “an administration and a military force which are purely personal instruments of the master.”

We see this pattern vividly illustrated by the Trump administration’s approach to staffing senior roles, including those leading high-stakes diplomatic negotiations.

Look at Steve Witkoff, a real estate developer and longtime Trump friend with no foreign policy experience, who has served as the administration’s lead envoy on some of the most sensitive negotiations in the world.

Or Jared Kushner, the president’s son-in-law, who, despite having no background in foreign policy, was entrusted with key roles in Middle East diplomacy, while his investment firm pursues deals with the same Gulf states he is negotiating with on behalf of his country.

Serving the sultan

These are not appointments that a merit-based system would produce. But right now in America, officials serve the sultan, not the republic, which is why their speeches are regularly given for an “audience of one.”

Furthermore, in seeking the sultan’s favour, appointees regularly debase themselves on television, such as when Kevin Warsh, Trump’s pick to be the next head of the Federal Reserve, refused to admit Trump lost the 2020 election.

This sultanistic pattern of rewarding loyalty and punishing defiance is expanding. Federal disaster relief, long treated as a non-partisan obligation of the government, has become a stark illustration of this logic.

Since the start of his second term, Trump has approved just 23 per cent of disaster funding requests from blue states, compared to 89 per cent for red states. In some cases, the conditionality for disaster aid has been made explicit: for example, in 2025, as fires ravaged Los Angeles, Trump threatened to withhold aid unless California enacted voter ID laws — a condition with no relationship to disaster recovery.

This fear of punishment also helps explain why, fearing for their businesses, many media companies are bowing to “the court of King Trump.”

‘Orgy of corruption’

Finally, Weber’s framework sheds light on what may be the most defining feature of the Trump administration: a blurring of the lines between public office and private enrichment. Under sultanism, the distinction between the ruler’s personal wealth and the state’s treasury is, at best, notional.

Trump and his team have governed accordingly, with perhaps the most egregious example being hundreds of millions of dollars of insider trading around the Iran war. In a healthy democracy, this “orgy of corruption” would be investigated and prosecuted. But in a patrimonial system this is simply how things work: the state exists to serve the ruler and his inner circle.

This is what the world must now manage. A sultanistic system does not respond to appeals to shared values or long-standing agreements. It responds to leverage, personal relationships with the ruler and transactional incentives.

Policymakers and business leaders increasingly understand they are dealing with a court that rewards fealty and punishes defiance. That’s why the Swiss gave Trump a gold bar in exchange for lower tariffs, and why the Qataris gave him a “palace in the sky.”

In 2026, appeals to shared democratic values or common national interests are pointless; bring the sultan something he wants or face punishment. Weber helps explain why.

The Conversation

Christopher Collins does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

ref. Donald Trump’s chaotic mess: When U.S. power serves the ‘sultan,’ global rules erode – https://theconversation.com/donald-trumps-chaotic-mess-when-u-s-power-serves-the-sultan-global-rules-erode-281941

Dissuasion algorithmique, rétention : l’IA fait-elle passer la cybersécurité à une nouvelle ère ?

Source: The Conversation – France in French (2) – By Christine Abdalla Mikhaeil, Assistant professor in information systems, IÉSEG School of Management

L’entreprise états-unienne Anthropic, spécialisée en intelligence artificielle générative, a pris début avril la décision de geler la diffusion publique de l’un de ses modèles récents, baptisé « Mythos ».

Selon un communiqué de la société, cette décision découle d’une puissance de calcul et d’une capacité de raisonnement jugées trop « offensives ». Anthropic a choisi de ne partager son modèle qu’avec une coalition de géants technologiques (Apple, Amazon Web Services, Cisco, Google, Microsoft, etc.) dans le cadre du projet Glasswing. Le but annoncé est d’utiliser Claude Mythos Preview pour détecter des vulnérabilités dites « zero-day » (c’est-à-dire inconnues et n’ayant aucun correctif connu) et sécuriser proactivement les logiciels critiques… avant que des acteurs malveillants n’exploitent ces failles.


Les grands modèles de langage savent déjà coder depuis quelques années, mais la presse spécialisée documente désormais un saut plus préoccupant. Des systèmes d’intelligence artificielle (IA) peuvent identifier des vulnérabilités réelles dans des logiciels critiques. Les autorités, comme l’Agence nationale de la sécurité des systèmes d’information (Anssi), soulignent la capacité des systèmes d’IA à automatiser les attaques.

Les enjeux de la diffusion massive de tels modèles, Mythos compris, dépassent largement le cadre technique. Une cyberattaque d’envergure, automatisée par une IA, pourrait paralyser des systèmes financiers ou logistiques en quelques secondes, avec un coût de remédiation se chiffrant en milliards d’euros. Les enjeux sont aussi sociétaux et de santé, puisque nos hôpitaux, nos réseaux énergétiques et les autres systèmes critiques reposent sur des couches logicielles souvent anciennes, vulnérables à des « attaques de zero-day » désormais générées à la chaîne.

Dans ce contexte, des IA ultraperformantes, comme Mythos, peuvent-elles contribuer à une forme de « dissuasion algorithmique » ? Celle-ci repose sur un principe simple : détecter et neutraliser ses propres vulnérabilités critiques plus vite que n’importe quel attaquant humain ou automatisé – y compris lors d’une attaque – si rapidement que l’attaque en devient inutile ou trop coûteuse.

La rétention de ce modèle par une ou des entreprises privées états-uniennes rouvre également la question de la souveraineté numérique à l’échelle mondiale.

Les systèmes d’IA facilitent les cyberattaques

Historiquement, la cybersécurité repose sur une asymétrie fondamentale : l’attaquant n’a besoin de trouver qu’une seule faille de sécurité, tandis que le défenseur doit toutes les combler dans une forme de course contre la montre.

L’intégration de systèmes d’IA renforce les capacités des attaquants, en premier lieu parce qu’un modèle comme Mythos peut scanner des millions de lignes de code en quelques minutes, là où un humain passait des semaines à analyser le code source d’un logiciel pour y déceler une erreur de mémoire. C’est ce que l’on appelle l’« automatisation de la reconnaissance ».

De plus, l’IA permet le phishing de haute précision, c’est-à-dire des messages frauduleux (le phishing classique) mais plus crédibles, sans fautes d’orthographe, dans n’importe quelle langue et ultrapersonnalisés pour tromper l’utilisateur. L’Anssi alerte d’ailleurs sur l’usage de l’IA générative pour briser les barrières linguistiques et psychologiques traditionnelles qui rend les « pare-feux humains » – c’est-à-dire la vigilance et l’esprit critique des lecteurs – de plus en plus obsolètes.

Enfin, des malwares peuvent désormais réécrire leur propre code en temps réel pour échapper à la détection « par signature ». Cette méthode classique des antivirus consiste à identifier un virus par son « empreinte » (un code déjà connu et répertorié). En changeant constamment de forme (exploitation polymorphe), le malware devient invisible pour ces outils traditionnels.

L’IA pour la cyberdéfense

En miroir, l’IA améliore aussi les capacités de cyberdéfense, grâce à des analyses causales, qui permettent de modéliser les relations entre événements, ainsi qu’en accélérant l’identification d’anomalies par une surveillance automatisée et la priorisation de leurs corrections. Ainsi, un système d’analyse IA a permis en janvier de découvrir 12 failles de sécurité dans OpenSSL, un logiciel essentiel à la protection des communications internet mondiales.

Mythos semble également déjà participer à cette automatisation, et Firefox affirme déjà avoir identifié et réparé 271 vulnérabilités grâce à ce logiciel, ce qui suggère que Mythos excelle effectivement dans la détection de vulnérabilités lorsqu’il a accès au code source.

En revanche, rien ne prouve pour l’instant que Mythos puisse, sans accès au code source et sans intervention humaine, compromettre de manière autonome n’importe quel logiciel fermé.

De plus, des analyses suggèrent que des capacités comparables seraient déjà reproductibles à partir de modèles publics, remettant en cause l’efficacité de cette rétention. Ainsi, Mythos ressemble aujourd’hui davantage à un analyste de sécurité surpuissant, capable d’identifier des failles et de proposer des pistes d’exploitation, qu’à une entité autonome de cyberattaque universelle.

Ce qui inquiète vraiment, ce n’est pas seulement que Mythos sache mieux coder ou mieux tester du code : c’est qu’il semble abaisser le coût, le temps et le niveau d’expertise nécessaires pour découvrir et enchaîner des failles, donc potentiellement accélérer aussi bien la défense que l’attaque.

Vers un nouvel équilibre de la « dissuasion algorithmique » ?

Dans ce contexte, la notion de « dissuasion algorithmique » (algorithmic deterrence en anglais) émerge. Elle peut être comprise par analogie avec la dissuasion nucléaire : il ne s’agirait plus seulement de se protéger, mais de posséder une capacité de réponse et de détection si rapide que l’attaque en devient inutile ou trop coûteuse.

Contrairement au domaine nucléaire, la dissuasion algorithmique repose sur le renforcement des mécanismes défensifs (plutôt que de réponse) : détection accélérée des intrusions, analyse des causes et simulation d’attaques pour boucher les failles avant qu’elles ne soient exploitées.

Avant l’ère de l’IA, la dissuasion algorithmique était plus limitée : les équipes de sécurité réalisaient des tests d’intrusion, une méthode d’évaluation de la sécurité qui repose sur la simulation de cyberattaques, pour identifier les ports ouverts et failles connues, pour enclencher leur correction.

Aujourd’hui, on l’a vu, l’IA peut renforcer les mécanismes défensifs et donc la dissuasion. Mais, dans le meilleur des cas, une IA de défense permet de réduire le coût de la protection et d’augmenter la probabilité qu’un attaquant soit repéré ou neutralisé avant d’atteindre son objectif.

La dissuasion algorithmique reste donc fragile. Même à l’ère de l’IA, elle dépend beaucoup des pratiques des acteurs opérationnels (agences de cybersécurité, armées, entreprises), de la qualité et de la modernité des systèmes hérités qu’ils doivent protéger et intégrer, des stratégies nationales et militaires mises en œuvre par les États ainsi que des dispositifs de gouvernance qui définissent les règles, les responsabilités et les mécanismes de contrôle de l’IA.

Le dilemme de la rétention d’outils de dissuasion : sécurité contre transparence

Ne pas rendre disponibles au grand public certains modèles peut sembler responsable, car cela évite de publier des capacités offensives qui seraient immédiatement détournées. Mais cette rétention concentre le pouvoir technologique entre quelques mains et réduit la transparence scientifique.

Le débat rejoint ici celui du cadre l’AI Act européen, qui impose déjà des obligations de transparence, de traçabilité et de documentation pour les modèles d’IA d’usage général, tout en cherchant à concilier innovation, sécurité et protection des secrets industriels.

L’opacité des modèles d’IA limite leur auditabilité, entrave le développement de contre-mesures appropriées et concentre le pouvoir technique entre quelques acteurs principalement américains, au détriment de la recherche ouverte et de la gouvernance démocratique. Cette critique s’inscrit dans une littérature académique plus large montrant que l’opacité des systèmes d’intelligence artificielle compromet leur reproductibilité, leur auditabilité et, in fine, leur valeur scientifique.

The Conversation

Christine Abdalla Mikhaeil est membre de Association for Information Systems (AIS). Elle a reçu des financements de l’Institut Bachelier et du LEM CNRS UMR 9221.

ref. Dissuasion algorithmique, rétention : l’IA fait-elle passer la cybersécurité à une nouvelle ère ? – https://theconversation.com/dissuasion-algorithmique-retention-lia-fait-elle-passer-la-cybersecurite-a-une-nouvelle-ere-282192

‘A life-and-death matter’: understanding how Ofsted inspections risk suicidal thoughts in teachers

Source: The Conversation – UK – By Rachel Harding, Research Fellow in Social Sciences, Nottingham Trent University

Zhuravlev Andrey/Shutterstock

Ofsted, the schools inspectorate in England, was the subject of a UK parliamentary inquiry after the death by suicide of Ruth Perry, headteacher of Caversham Primary school in Berkshire, in 2023. The coroner’s report had concluded that Perry’s death was “suicide, contributed to by an Ofsted inspection”.

The parliamentary inquiry called for submissions of evidence about Ofsted from members of the public. Our recent research has analysed the 233 published submissions, many of which were from teachers. One submission to the inquiry included an impact statement by a headteacher written in 2022. It read:

The manner in which the inspection was conducted and the lack of integrity from the Lead Inspector has meant that my family have had to support me through suicidal thoughts and through countless occasions of being in floods of tears as soon as I think back to that day.

“It seems incredible that an issue like the conduct of school inspection should be a life-and-death matter, but so indeed it has become,” the submission from her school stated.

Theory of suicide

Sociological theory helps us ask questions and seek radical answers about how societies function, including government policy such as the inspection of schools.

Sociologist Émile Durkheim’s theory of suicide argues that suicide does not only happen because of mental illness, but that it also has a social context. Durkheim examined how the interaction of people and social control, as well as notions of shame, guilt, failing expectations and feeling trapped, might result in someone having suicidal thoughts and feelings.

We found evidence of teachers feeling shame. One submission mentioned “the enormous shame and distress that is felt by those leading and working within the school”.

Teachers also reported feeling trapped:

In my last inspection in November 2019, I lost half a stone in the three days (starting from the phone call) and lost my voice. My family suffered, there were arguments and I slept on the couch. The stress and pressure was all too much. As a school leader, I live in fear and I came into education because I love teaching but now I feel trapped.

man sat alone in classroom
The impact of Ofsted inspections on teacher wellbeing is well documented.
Elnur/Shutterstock

The risk of a less than good inspection was “petrifying”. Having to be always ready for an inspection was “intolerable”. The thought of letting colleagues down by making a mistake was “unbearable”.

Teachers wrote about ill health because of Ofsted experiences. These accounts included vomiting, physical collapse, panic attacks, incontinence and suspected stroke with a temporary loss of speech. One wrote that they had a miscarriage the day after a deeply stressful Ofsted inspection.

The government and Ofsted’s response

The Education Committee’s report noted that the committee had heard that “Ofsted has lost trust and credibility among many in the teaching profession.”

However, a number of reports on Ofsted’s practice, including the independent learning review commissioned by Ofsted, fail to acknowledge that teachers can have suicidal thoughts and feelings because of Ofsted.

Ofsted’s developments since the inquiry include introducing report cards for schools. Ofsted says this is fairer, but teachers say it creates more stress. An independent risk assessment warns that “the revised framework does not reduce the pressure on leaders to achieve a desirable outcome. The consequence of not meeting the expected standards of the revised framework will remain high stakes in nature.”

Other developments include changes regarding inspections of provision for children in care and inspection frameworks themselves.

But we do not believe that these changes constitute the “root and branch” review of Ofsted previously called for by education leaders.

Professor Julia Waters, Ruth Perry’s sister, has said that our study “presents the evidence of the terrible human cost posed by Ofsted inspections, evidence that Ofsted and successive governments have still not fully grasped”.

Both Ofsted and the government should review how the inspectorate works. Not to do so runs the risk of school inspections remaining a life-and-death matter.

If you’re struggling with suicidal thoughts, the following services can provide you with support:
In the UK and Ireland – call Samaritans UK at 116 123.
In the US – call the National Suicide Prevention Lifeline at 1-800-273-TALK (8255) or IMAlive at 1-800-784-2433.
In Australia – call Lifeline Australia at 13 11 14.
In other countries – visit IASP or Suicide.org to find a helpline in your country.

The Conversation

The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

ref. ‘A life-and-death matter’: understanding how Ofsted inspections risk suicidal thoughts in teachers – https://theconversation.com/a-life-and-death-matter-understanding-how-ofsted-inspections-risk-suicidal-thoughts-in-teachers-278478

AI in the emergency department: promising, powerful but still unproven

Source: The Conversation – UK – By Ewen Harrison, Professor of Surgery and Data Science, University of Edinburgh

Gorodenkoff/Shutterstock.com

Artificial intelligence can now outperform doctors at diagnosing patients in the emergency department, according to a new study in Science.

The AI was given written notes from real emergency department records from a hospital in Boston, US, and asked to weigh in at different points during the patient’s care. At the earliest stage – triage, when a patient first arrives – the AI identified the correct diagnosis, or something closely related, in 67% of cases.

The two doctors used for comparison managed 50% and 55%. That’s a meaningful gap, especially at the moment when information is scarcest and uncertainty is highest.

This study matters because the field is moving so fast. Earlier research showed that large language models – the technology behind systems like ChatGPT – could pass medical licensing exams. Interesting, but not all that illuminating. Passing an exam is not the same as being useful on a ward.

This new study goes further. It puts AI alongside doctors across several tasks, using genuine clinical text from a real emergency department. That makes it more directly relevant to medical practice than most of what’s come before. It suggests these systems are developing into something that could genuinely help doctors think through a wide range of possible diagnoses, especially in situations where missing a serious condition is the main concern.

There are good reasons, though, not to get carried away.

The AI was working entirely from written text. It never saw the patient, never noticed how breathless or frightened they looked, never examined them, spoke to their family, weighed up the chaos of a busy department, or took any responsibility for what happened next. It was not practising emergency medicine. It was offering a written opinion based on selected information.

There’s also a gap between producing a list of possible diagnoses and actually improving patient outcomes. A longer list might help a doctor think more broadly, but it could equally generate new problems: unnecessary tests, over-treatment, extra workload, or unwarranted confidence in an answer that sounds plausible but turns out to be wrong.

And some of the benchmark cases used in studies like this may have been publicly available when the AI was trained, which doesn’t undermine the emergency department findings, but is another reason to treat headline numbers with some scepticism.

The hard question

So the question isn’t really whether AI can help doctors think through difficult cases. The harder question is how this should be tested and governed in real clinical settings like the NHS.

That question is already urgent. A Royal College of Physicians snapshot found that 16% of UK doctors were using AI tools in clinical practice every day, with another 15% doing so weekly. Doctors are already using these tools in their daily work – before hospitals and health systems have properly worked out how to assess them, train staff to use them safely, spot when they’re causing harm, or decide who is responsible when something goes wrong.

A doctor looking at a tablet computer.
Around 16% of doctors in the UK use AI every day.
Josep Suria/Shutterstock.com

It’s tempting to say that the solution is to keep a human in the loop. But that phrase does very little work on its own. We need to know which human, in which loop, and with what authority. A doctor’s ability to override an AI suggestion is not, by itself, a safety system. Someone still has to decide which tools get used, who can change how they behave, how harms are spotted, and who is responsible when the tool quietly starts failing.

This study represents genuine progress. But it doesn’t, on its own, change how medicine should be practised. The right response is neither to prohibit these systems nor to let them quietly become part of the routine before anyone has thought it through. They should be trialled in real clinical settings, used as a form of second-opinion support rather than a substitute for clinical judgment, and measured against what actually matters to patients: care that is better, safer and faster.

The Conversation

Ewen Harrison receives funding from a number of grant-giving bodies including UKRI, NIHR, HDRUK, and Wellcome Leap. He is a Deputy Editor with NEJM AI.

ref. AI in the emergency department: promising, powerful but still unproven – https://theconversation.com/ai-in-the-emergency-department-promising-powerful-but-still-unproven-282029

What Iran’s absence from the Venice Biennale reveals about art and politics

Source: The Conversation – UK – By Katayoun Shahandeh, Lecturer in Museum Studies, SOAS, University of London

Just days before the opening of the 2026 Venice Biennale, organisers announced that Iran would no longer participate.

A short statement posted to the Venice Biennale website on May 4 said: “With regard to the National Participations in the 61st International Art Exhibition…it has been announced that the Islamic Republic of Iran will not participate.” No explanation was given. I believe that silence is itself revealing.

Iran’s withdrawal is less a sudden decision than the result of converging geopolitical and economic pressures that are reshaping both the global art world and Iran’s place within it.

At the most immediate level, the withdrawal reflects the material realities of crisis. With internet access restricted, international flights suspended and communication networks severely disrupted, even the basic logistics of participation – coordinating, shipping and installing artworks – probably became nearly impossible for Iran.

These conditions have been compounded by intensifying economic pressures, including the sharp devaluation of the Iranian rial, which has made international cultural engagement increasingly difficult to sustain.

An explanation of the Venice Biennale.

Such constraints point to a fundamental condition of contemporary art: global exhibitions rely on infrastructures of mobility and communication that are easily destabilised by conflict and sanctions.

The timing is also significant. The decision comes amid renewed military tensions and escalating political rhetoric surrounding Iran’s position in the global order. In such moments, when political discourse edges toward existential threat, the stakes of cultural visibility are heightened. At the same time, sustaining cultural presence becomes more difficult.




Read more:
Middle East conflict looks increasingly like a war nobody can win


More revealing still was the lack of any announced artist, curatorial framework or exhibition concept for Iran’s pavilion, even days before the Biennale’s opening.

Iran’s presence at the Venice Biennale has historically been organised through state institutions, with oversight exercised by the ministry of culture and Islamic guidance since the Iranian revolution (1978-79). As with many national pavilions, this model positions art as a form of cultural diplomacy. But in Iran’s case, it has often produced a disconnect between official representation and contemporary artistic practice.

This gap is significant. The Venice Biennale, often described as the “Olympics of the art world”, remains structured around national pavilions, with each country responsible for presenting its cultural identity on a global stage. Yet, as critics have long argued, it has never been a neutral platform, but a space where art and geopolitics intersect.

More broadly, biennials are deeply embedded in political and institutional contexts, rather than existing outside them. Within this framework, they are often understood as sites of cultural soft power, where nations project influence through artistic production.

National representation in crisis

Iran’s withdrawal must also be understood in relation to the wider turmoil surrounding the 2026 biennale itself. This year’s edition has been marked by extraordinary controversy, including disputes over the involvement of Russia and Israel, calls for boycotts and the resignation of the entire international jury just days before the opening.

These events expose the fragility of the biennale’s longstanding claim to neutrality. Rather than existing outside politics, it has become a site where geopolitical tensions are actively staged and contested.

To exhibit at the biennale is never neutral: it means entering a highly visible arena shaped by competing narratives of legitimacy and power. For the Islamic Republic, this raises a deeper tension. The biennale’s national pavilion model requires countries to present a coherent cultural identity through contemporary art. Yet Iran’s artistic landscape is anything but singular. It is shaped by internal contradictions between state and independent practices, censorship and experimentation and local production and diasporic circulation.

The entire jury resigned just days before the opening.

These tensions are difficult to reconcile within a state-managed exhibition framework. The very premise of the pavilion – art as national representation – sits uneasily with a system in which artistic expression is subject to ideological and institutional control.

At the same time, the Biennale embodies forms of global circulation, cultural competition and visibility tied to international art markets that do not always align with the cultural and political ethos of the Islamic Republic. Representation therefore involves negotiating how a nation appears, to whom, and on whose terms.

The current moment makes this tension even more acute. As political rhetoric escalates and the possibility of large-scale destruction is invoked in global discourse, cultural visibility becomes more urgent. Art offers one of the few spaces through which narratives beyond conflict and diplomacy can emerge. Yet for Iranian artists, cultural presence is becoming more fragmented, shaped by diasporic networks, constrained by national borders and limited by economic and infrastructural pressures.

Iranian artists, particularly those working through independent and diasporic networks, have for decades operated beyond the frameworks of state representation, with their work circulating internationally through alternative artistic circuits. Iran’s missing pavilion, then, does not signal the disappearance of Iranian art. Rather, it reveals the precarious conditions through which that art circulates.

Iran’s absence from the Venice Biennale also highlights the limits of the national pavilion model. The system has frequently been criticised for reducing complex artistic practices to simplified national identities, even as contemporary art now operates through transnational networks that exceed the boundaries of the nation-state.

In Venice this year, the missing pavilion reflects an art world shaped as much by political crisis as by artistic production. Iranian art is not absent from the global stage. Yet the conditions under which it circulates and remains visible have become increasingly fragile.

The Conversation

Katayoun Shahandeh works for SOAS University of London.

ref. What Iran’s absence from the Venice Biennale reveals about art and politics – https://theconversation.com/what-irans-absence-from-the-venice-biennale-reveals-about-art-and-politics-282416

Motown’s Black women songwriters and producers were the invisible architects behind the pop music juggernaut

Source: The Conversation – USA (2) – By Margena A. Christian, Emeritus professor, University of Illinois Chicago

Sylvia Moy was a trailblazing Motown songwriter. L. Busacca/WireImage for Songwriter’s Hall of Fame

During the 1960s, in a country divided by racial strife, the music of Berry Gordy Jr.’s Motown Records helped bring people together.

Motown was noted for star performers like Mary Wells, The Miracles, The Supremes, The Temptations, Martha Reeves and the Vandellas, Marvin Gaye and Stevie Wonder. But, behind the scenes, a talented group of lesser known women were driving the hits in Hitsville U.S.A.

I’m a scholar of popular culture and author of the biography “It’s No Wonder: The Life and Times of Motown’s Legendary Songwriter Sylvia Moy.” Researching my book inspired me to find other women who contributed to the Detroit label’s era of chart dominance and helped change the music industry, despite going largely unrecognized for their efforts.

I listened to Motown growing up, but it wasn’t until 2021, while sitting at home during the pandemic, that I discovered Moy’s history as the lyricist for Stevie Wonder and how she helped revive his early career.

Because Moy died in 2017, I wasn’t able to speak with her for the book. Instead I researched her life by reading countless interviews she gave, along with talking to her former colleagues at Motown, family and ethnomusicologists, who are scholars that study music through the lens of culture.

Architect of the early sound

When Gordy was organizing his company, Janie Bradford was one of the original five founding members who arrived in 1958. She was the label’s first secretary and its first female songwriter after co-writing, with Gordy, the song “Money (That’s What I Want).” That song was released on Tamla Records in 1959 and performed by Barrett Strong. When Motown was incorporated the following year, the song became the label’s first hit record on the R&B chart and Billboard Hot 100.

Woman stands behind a podium and speaks into a microphone.
Janie Bradford speaks at a 2022 tribute to Mary Wilson, a member of The Supremes, in Los Angeles. Bradford was one of the founding members of Motown.
Alison Buck/Getty Images for The Recording Academy

Later, Bradford co-wrote “Contract on Love” for Wonder and “Too Busy Thinking About My Baby,” first recorded by The Temptations and later, Marvin Gaye. Bradford, who later became Motown’s director of writer’s relations, teamed up with pianist Richard “Popcorn” Wylie in the early 1960s to form Janard, a small production company.

Bradford’s collection of poetry is what captured Gordy’s attention, so he encouraged her to be a songwriter. Her witty lyrics told stories about situations that most anyone could relate to – namely, money and love – blended with up-tempo, thumping beats.

Laying the foundation as a producer

Another key figure who paved the way with the Motown sound was Raynoma Gordy Singleton, who was married to Berry Gordy Jr. from 1960 to 1964. She organized Motown during its beginnings by completing the necessary paperwork to incorporate the business. Known as “Miss Ray” to some and “Mother Motown” to others, she located the legendary house at 2648 West Grand Boulevard that became the Motown headquarters and, decades later, the Motown Museum.

In her role as the label’s first executive vice president, she established a tape library. A piano virtuoso and singer, the Cass Technical High School graduate wrote that she was able to play all string and wind instruments. As a result, she became the company’s first female arranger and producer by putting together its first backup vocal group, the Rayber Voices, in 1958.

“Producing records was where the action was controlled – and where the money was to be made,” she wrote in her memoir, “The Untold Story: Berry, Me, and Motown,” which aimed to reclaim her place in the Motown echelon.

During the 1960s, women weren’t considered producers because of broader biases and norms in the male-dominated music industry. Even so, Miss Ray got credit for producing Jimmy Ruffin’s song “Don’t Feel Sorry for Me” in 1961.

Earning a producer’s credit was a sign of legitimacy. Most producers received a songwriting credit and determined who received credit in the liner notes for their contribution to the recording.

While women mostly worked in administrative roles at Motown, there still weren’t any female full-time, in-house songwriters and producers. Like the rest of the music industry back then, Motown’s internal structure was patriarchal with those positions.

The first certified female songwriter and producer

Yet this imbalanced gender dynamic at Motown didn’t stop Sylvia Moy.

There hadn’t been any women producers behind significant, popular songs at Motown until Moy arrived, according to interviews I conducted for her biography.

Motown was at its peak in 1964. Demand for new songs was intense. When the label’s executives realized how skillfully the two audition songs Moy performed were composed, they decided that her future was in songwriting instead of singing.

Discovered by William “Mickey” Stevenson and Marvin Gaye, Moy was hired as the first female in-house songwriter, competing with eminent colleagues like Smokey Robinson, Norman Whitfield and the songwriting trio Holland-Dozier-Holland who wrote 10 of the Supremes’ chart-topping singles. Moy made more history in 1965 after co-writing and co-producing Stevie Wonder’s “Uptight (Everything’s Alright).”

While she received the songwriting credit and helped revive the teenaged Wonder’s career, Moy wasn’t given the producer’s credit, unlike her two male counterparts, Stevenson and Henry “Hank” Cosby.

A lack of recognition stymied Moy’s career opportunities. If a songwriter or producer wasn’t credited, their value could not be validated or established, which made it harder for them to find work at other record labels.

According to my research, Moy revealed that she never got producer credit for any of her work while at Motown. This is why her legacy was buried for so long.

Other tunes she wrote for Wonder were “I Was Made to Love Her,” “My Cherie Amour” and “With A Child’s Heart,” co-written with Vicki Basemore. Moy also wrote Marvin Gaye and Kim Weston’s “It Takes Two” and The Isley Brothers’ “This Old Heart of Mine (Is Weak For You).” Though songwriter Eddie Holland told me he gave her a co-writing credit for “This Old Heart of Mine,” Moy’s name was not listed on the record, only Holland-Dozier-Holland.

Interviews I conducted with Moy’s family members and research from an ethnomusicologist suggest she was even an uncredited co-writer for Wonder’s “Signed, Sealed, Delivered (I’m Yours),” his first song as a solo producer, and The Temptations’ “Ain’t Too Proud to Beg.”

However, Holland denied this claim in an interview with me, though he also admitted that the song’s late co-writer and producer, Norman Whitfield, presented him with the lyrics, and he wasn’t sure where they came from.

Full credit along with creative control

In 1968, Valerie Simpson became Motown’s first female songwriter to also receive a producer credit. This possibly happened because her songwriting partner was her husband, Nickolas Ashford.

Other famous female songwriters like Carole King, Ellie Greenwich and Cynthia Weil also had a prominent husband in the music industry. Sylvia Moy did not, which made what she did unprecedented.

A man and a woman stand for a portrait
Valerie Simpson poses next to her husband, Nickolas Ashford. Together, they formed the famed singing and songwriting duo Ashford and Simpson. She was the first woman songwriter and producer at Motown to receive complete credit for her creative contributions.
Aaron Rapoport/Corbis/Getty Images

Simpson told Billboard in 2023 that the credit was difficult to attain because so few women were producers back then. It finally happened with the Tammi Terrell and Marvin Gaye song “Ain’t Nothing Like The Real Thing,” with Simpson getting credit for co-writing, co-producing and performing background vocals along with Ashford.

This was their third hit tune by Terrell and Gaye, who also recorded “Ain’t No Mountain High Enough” and “Your Precious Love,” in 1967. The following year, they had another hit with “You’re All I Need to Get By,” which Ashford and Simpson also co-wrote, co-produced and did background vocals on.

‘Ain’t Nothing Like the Real Thing’ was performed by Marvin Gaye and Tammi Terrell. Valerie Simpson co-wrote and co-produced the song along with her husband, Nickolas Ashford.

Simpson became the first Black woman to be inducted into the Songwriters Hall of Fame in 2002. Moy became the second in 2006.

Though female songwriters and producers continue the fight for inclusion in the recording studio, the doors were opened by the tenacious women of Motown. It is because of them that future generations of female creatives know what is possible.

The Conversation

Margena A. Christian does not work for, consult, own shares in or receive funding from any company or organization that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

ref. Motown’s Black women songwriters and producers were the invisible architects behind the pop music juggernaut – https://theconversation.com/motowns-black-women-songwriters-and-producers-were-the-invisible-architects-behind-the-pop-music-juggernaut-278514

Lower East Side street named for ‘King of Comics’ Jack Kirby, a nod to one of the countless kids of immigrants who shaped the genre

Source: The Conversation – USA (2) – By Miriam Eve Mora, Managing Director of the Raoul Wallenberg Institute, University of Michigan

The Thing is from the fictional Yancy Street in Manhattan’s Lower East Side, where creator Jack Kirby was raised. Richie S/flickr, CC BY

The gesture may lack the explosive drama of a rooftop fight or the tension of a car chase, but on May 11, 2026, a street sign honoring a legendary comics creator will be unveiled in Manhattan’s Lower East Side.

After a lobbying effort by comics expert Roy Schwartz, the New York City Council in December 2025 approved the naming of a block of Essex Street between Delancey and Rivington streets in honor of Jack Kirby.

Black-and-white photo of middle-aged white man smoking a pipe.
Comic book artist Jack Kirby attends San Diego Comic Con in 1973.
Clay Geerdes/Getty Images

Kirby, born Jacob Kurtzberg in 1917 to Jewish immigrants, spent roughly the first 40 years of his life in New York, aside from a stint serving in the military during World War II. Before enlisting, he’d already embarked on a career as a comics artist. He went on to become a key figure during the medium’s golden age, a period that most scholars and fans agree began with the creation of Superman in 1938 and ended with the implementation of the Comics Code Authority in 1956, which heavily restricted content until enforcement weakened in the 1970s.

Though you may not have heard of Kirby, you’d have to deliberately avoid pop culture to miss his most influential creations: Captain America, the Fantastic Four, X-Men, Thor, Hulk, Iron Man and Black Panther.

For my part, however, as a scholar of American Jewish immigration history – and as a lifelong comic book fan – I hold a place of reverence for the man known as the “King of Comics.”

Jewish American history, immigration history, the history of New York City and the origins of the comics industry are inextricably linked. New York played a starring role in the golden age of comics. And like Kirby, many of the genre’s most famous artists were Jewish.

Jewish immigrants put pen and ink to paper

Comics found a wide audience in New York City during their early years in the late 19th and early 20th centuries, from early newspaper strips like “The Yellow Kid” and “Abie the Agent” to later ones like “Little Orphan Annie.”

As World War II drew to a close in the summer of 1945, there was a citywide newspaper delivery strike, leaving many New Yorkers desperate for news and entertainment – so much so that Mayor Fiorello LaGuardia took it upon himself to read the Sunday comic strips over the radio, performing them with characteristic vigor and enthusiasm.

Among the first publications that would today be recognizable as “comic books” were compilations of these early newspaper strips, assembled by newsprint salesman and Jewish New Yorker Max Gaines. Gaines, born Maxwell Ginzburg, compiled various comic strips into neatly packaged, inexpensive entertainment for the masses, helping pioneer the saddle-stitched comic book – thin, stapled magazines that would become the primary format for superhero stories.

As the superhero genre took off in the late 1930s, other publishers emerged from Jewish New York. Harry Donenfeld and Jack Leibowitz, in partnership with Malcolm Wheeler-Nicholson, created Detective Comics and Action Comics, which helped establish the company later known as DC Comics.

In addition to early publishers, many pioneering comics artists were raised in New York City as the children of Jewish immigrants, including Marvel Universe architect Stan Lee and his brother, Larry Lieber; Will Eisner, creator of “The Spirit” and co-creator of “Sheena: Queen of the Jungle”; and Al Jaffee, a longtime contributor to Mad Magazine.

An ode to the Lower East Side

In Jack Kirby’s comics, the city shines through.

The Fantastic Four – the superhero squad that Kirby created with Stan Lee – operates out of midtown Manhattan’s fictional Baxter Building, which Kirby modeled after the city’s mid-century skyscrapers.

Kirby also based the character of Ben Grimm – The Thing – on himself, mining his own life to write Grimm’s backstory. Grimm’s home is on the fictional Yancy Street, a tribute to Kirby’s own working-class upbringing on the Lower East Side’s Delancey Street. The thoroughfare is rich with Jewish history and in close proximity to iconic businesses like Katz’s Deli and Russ and Daughters.

Another of Kirby’s most iconic characters was Steve Rogers – Captain America – which he co-created with Joe Simon.

A poor orphan from Brooklyn, Rogers attempts to enlist in the U.S. Army to fight the Axis powers during World War II, but is rejected as unfit for duty. He is later recruited into Project Rebirth, where he is transformed into a super-soldier after being injected with a serum designed to maximize human physical and mental abilities.

Captain America attracted legions of fans among American youth, many of whom saw themselves in the superhero. Though Rogers is Christian, his story of transformation from weakling to hero certainly spoke to young Jewish boys and men, who were often inaccurately portrayed in the media and press as intellectually superior but physically inferior.

Captain America, though fictional, is already recognized as a part of New York City history, and has a statue in Brooklyn, which was unveiled in 2016 with the inscription “I’m just a kid from Brooklyn.”

The city as a muse

Even comics created by artists outside New York City – like Ohio natives and Superman co-creators Jerry Siegel and Joe Shuster – are, by virtue of their content, still in many ways New York comics.

The glittering Metropolis in “Superman” is widely understood as a stand-in for New York; for example, in the April 1950 issue of Action Comics, the Statue of Liberty is said to appear in “Metropolis Harbor.”

A bronze statue of a muscular superhero who's hoisting a shield with a star on it into the air.
A Captain America statue is unveiled during a ceremony at Prospect Park in New York’s Brooklyn borough on Aug. 10, 2016, in honor of the character’s 75th anniversary.
Angela Weiss/AFP via Getty Images

If Metropolis is the bright, shining, optimistic view of the city, then Gotham, the home of Batman, reprises the city through a grittier lens.

Writer Washington Irving had first described New York as Gotham in the early 1800s. But by the time Batman came on the scene, the term had become less common in everyday speech, and DC Comics repurposed the name for the fictional Gotham City. Beyond the name, Gotham City’s architecture, bridges, boroughs and neighborhoods are an homage to New York.

By officially recognizing Jack Kirby, the city adds the artist to a distinguished roster of politicians, community activists and celebrities honored with street names.

Jack Kirby Way celebrates a legendary comics artist while also acknowledging the immigrant creators who helped shape the genre. It’s a fitting tribute: As much as the comics industry is indebted to the city, the city is indebted to the comics industry.

The Conversation

Miriam Eve Mora does not work for, consult, own shares in or receive funding from any company or organization that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

ref. Lower East Side street named for ‘King of Comics’ Jack Kirby, a nod to one of the countless kids of immigrants who shaped the genre – https://theconversation.com/lower-east-side-street-named-for-king-of-comics-jack-kirby-a-nod-to-one-of-the-countless-kids-of-immigrants-who-shaped-the-genre-279716