Sony AI's autonomous table-tennis robot Ace has become the first robot to compete against top-level human players. Reuters reports: Ace, created by the Japanese company Sony's AI research division, is the first robot to attain expert-level performance in a competitive physical sport, one that requires rapid decisions and precision execution, the project's leader said. Ace did so by employing high-speed perception, AI-based control and a state-of-the-art robotic system. There have been various ping-pong-playing robots since 1983, but until now they were unable to rival highly skilled human competitors. Ace changed that with its performances against human elite-level and professional players in matches following the rules of the International Table Tennis Federation, the sport's governing body, and officiated by licensed umpires. The project's goal was not only to compete at table tennis but to develop insights into how robots can perceive, plan and act with human-like speed and precision in dynamic environments. In matches detailed in the study, Ace in April 2025 won three out of five versus elite players and lost two matches against professional players, the top skill level in the sport. Sony AI said that since then Ace beat professional players in December 2025 and last month. "The success of Ace, with its perception system and learning-based control algorithm, suggests that similar techniques could be applied to other areas requiring fast, real-time control and human interaction -- such as manufacturing and service robotics, as well as applications across sports, entertainment and safety-critical physical domains," said Peter Durr, director of Sony AI Zurich and leader for Sony AI's project Ace. The findings have been published in the journal Nature.

Read more of this story at Slashdot.

Derrière les notes de mise à jour en apparence banales de Firefox 150 se cache un véritable séisme pour la cybersécurité. En s'alliant avec la nouvelle IA d'Anthropic, Mozilla a débusqué et corrigé près de 300 failles d'un coup. Une avancée historique qui pourrait bien signer la fin des attaques « zero-day » et définitivement inverser le rapport de force entre pirates et défenseurs.

Selon Bloomberg, un petit groupe d’utilisateurs issus d’un groupe Discord aurait eu accès à Claude Mythos, le modèle hyper‑restreint d’Anthropic, contournant ainsi les protections du modèle d'IA le plus scruté du moment.

Derrière les notes de mise à jour en apparence banales de Firefox 150 se cache un véritable séisme pour la cybersécurité. En s'alliant avec la nouvelle IA d'Anthropic, Mozilla a débusqué et corrigé près de 300 failles d'un coup. Une avancée historique qui pourrait bien signer la fin des attaques « zero-day » et définitivement inverser le rapport de force entre pirates et défenseurs.

Il fut un temps où pour interagir avec une intelligence artificielle, on ouvrait naturellement ChatGPT. Mais en 2026, le chatbot d'OpenAI n'est plus seul sur son trône, et ses concurrents ont définitivement cessé de faire de la figuration. Voici les meilleures alternatives à utiliser.

Plusieurs outils d'intelligence artificielle générative rencontrent des soucis de fonctionnement en ce lundi 20 avril 2026.

An anonymous reader quotes a report from TechCrunch: The winning runner at a Beijing half-marathon for humanoid robots finished the race today in 50 minutes and 26 seconds -- significantly faster than the human world record of 57 minutes recently set by Jacob Kiplimo. [...] [T]he winning time is a massive improvement over last year's race, when the fastest robot finished in two hours and 40 minutes. The Associated Press reports that this year's winner was built by Chinese smartphone maker Honor. It seems the winning robot wasn't actually the fastest, as a different Honor robot finished in 48 minutes and 19 seconds. But that one was remote controlled -- the 50:26 robot was autonomous and won due to weighted scoring. About 40% of participating robots competed autonomously, while the remaining 60% were remote controlled, according to Beijing's E-Town tech hub. Not all of them did as well as Honor's robots, with one robot falling at the starting line and another hitting a barrier.

Read more of this story at Slashdot.

Vous cherchez un jeu à lancer sur votre PS5 mais vous pensez avoir épuisé toutes les idées ? Numerama a regroupé dans cette liste nos coups de cœur de la console de Sony que vous soyez sur Slim ou Pro : une liste de 20 jeux du survival horror au FPS nerveux jusqu'au RPG.

Boston Dynamics has integrated Google DeepMind into its robotic dog Spot, giving it more autonomous reasoning for industrial inspections like spotting spills and reading gauges. Spot can also now recognize when to call on other AI tools. IEEE Spectrum reports: Boston Dynamics is one of the few companies to commercially deploy legged robots at any appreciable scale; there are now several thousand hard at work. Today the company is announcing that its quadruped robot Spot is now equipped with Google DeepMind's Gemini Robotics-ER 1.6, a high-level embodied reasoning model that brings usability and intelligence to complex tasks. [T]he focus of this partnership is on one of the very few applications where legged robots have proven themselves to be commercially viable: inspection. That is, wandering around industrial facilities, checking to make sure that nothing is imminently exploding. With the new AI onboard, Spot is now able to autonomously look for dangerous debris or spills, read complex gauges and sight glasses, and call on tools like vision-language-action models when it needs help understanding what's going on in the environment around it. "Advances like Gemini Robotics-ER 1.6 mark an important step toward robots that can better understand and operate in the physical world," Marco da Silva, vice president and general manager of Spot at Boston Dynamics, says in a press release. "Capabilities like instrument reading and more reliable task reasoning will enable Spot to see, understand, and react to real-world challenges completely autonomously." You can watch a demo of Spot's new capabilities on YouTube.

Read more of this story at Slashdot.

La bixonimanie n’existe pas, mais des articles de blogs et des prépublications sur cette maladie inventée par une chercheuse en médecine ont suffi à tromper plusieurs intelligences artificielles conversationnelles et à s’introduire dans un article scientifique officiel, soulevant d'inquiétantes questions sur la vérification des données à l’ère de la démocratisation de l’IA.

[Deal du jour] Que faisiez-vous en 2024 ? Eh bien, pendant ce temps, Team Asobi, le studio de développement japonais, sortait ce qui est aujourd'hui considéré comme le « Mario de la PS5 ». Sacré GOTY 2024, le jeu se trouve rarement en promotion, sauf aujourd'hui sur Amazon.

Le 2 avril 2026, Anthropic a affirmé avoir identifié, au cœur de son modèle Claude, des « représentations internes » proches d’émotions. Une découverte qui éclaire pourquoi les IA semblent parfois empathiques… et comment ces états influencent réellement leurs décisions.

Sorti quelques mois plus tôt outre-Atlantique, le mini frigo à l’effigie d’Astro Bot est arrivé en France le 14 mars 2026. Mais difficile de mettre la main dessus : après une rupture temporaire, l’accessoire est de nouveau en stock sur Micromania.

Coup dur pour Anthropic. L'entreprise derrière le très performant modèle Claude a accidentellement rendu public le code source de son outil pour développeurs, Claude Code. Des internautes ont fouillé dans les dossiers et ont découvert plusieurs projets de l'entreprise. Y compris un curieux concept de Tamagotchi.

Coup dur pour Anthropic. L'entreprise derrière le très performant modèle Claude a accidentellement rendu public le code source de son outil pour développeurs, Claude Code. Des internautes ont fouillé dans les dossiers et ont découvert plusieurs projets de l'entreprise. Y compris un curieux concept de Tamagotchi.

Utility-scale solar construction... by robots! It's "one of the largest real-world demonstrations," notes Electrek, with 100 MW of capacity installed by the "Maximo" robots from AES, one of the world's top power companies. Maximo uses AI "to automate the heavy lifting of solar panels and accelerate solar installation," according to their web page, which shows a video of Maximo at work installing a vast field of solar panels in Kern County, California. With assistance from Nvidia, the Maximo team could "develop, test and refine robotic capabilities through physics-based simulation and AI driven modeling before deploying updates in the field," reports Electrek, and they're aiming for a full GW of solar generating capacity: After completing the first half of the Bellefield complex last summer, Maximo engineers went into a higher gear, with the latest version 3.0 robots consistently surpassing an installation rate of one module per minute, with construction crews installing as many as 24 solar panel modules per hour, per person. If that sounds fast, that's because it is. At full tilt, the latest Maximo robot-equipped crews have nearly doubled the output of traditional installation methods at similar solar locations throughout Southern California. "Reaching 100 MW is an important milestone for Maximo and for the role robotics can play in solar construction," explains Chris Shelton, president of Maximo. "It demonstrates that field robotics can move beyond experimentation and deliver consistent results at utility scale. As solar deployment continues to accelerate globally, technologies that improve installation speed, quality and reliability will become increasingly important...." Like just about every other business that demands a high degree of physical labor, the construction industry is facing huge labor shortages, making machines like Maximo that provide real efficiency gains welcome additions to the job site. "The combination of AI, vision, robotics and simulation driven engineering reduced development and validation timelines," the Maximo team said in a statement, "and increased confidence in field performance as the robotic fleet scaled."

Read more of this story at Slashdot.

Longtime Slashdot reader theodp writes: In Melania and the Robot, the New York Times reports on First Lady Melania Trump's inaugural Fostering the Future Together Coalition Summit, which brought together international leaders, First Spouses from around the world, tech leaders, educators, and nonprofits to collaborate on practical solutions that expand access to educational tools while strengthening protections for children in digital environments (Day 2 WH summary). The Times begins: "On Wednesday, Mrs. Trump appeared at the White House alongside Figure 3, a humanoid, A.I.-powered robot whose uses, according to the company that makes it, include fetching towels, carrying groceries and serving champagne. But Mrs. Trump joins tech executives and some researchers in envisioning a world beyond robot butlery. She is interested in how these robots could cut it as educators. Both clad in shades of white, the first lady and the visiting robot walked into a gathering of first spouses from around the world, a group that included Sara Netanyahu of Israel, Olena Zelenska of Ukraine, and Brigitte Macron of France. The dulcet tones from a (presumably human) military orchestra played as the first lady and her guest entered the event. Both lady and robot extolled the virtues of further integrating robots into the educational and social lives of children. In the history of modern first-lady initiatives, which have included building a national book festival (Laura Bush), reshuffling the food pyramid (Michelle Obama) and advocating for free community college (Jill Biden), Mrs. Trump's involvement of a humanoid robot in education policy was a first." "Figure 3 delivered brief remarks and delivered salutations in several languages. With its sleek black-and-white appearance, Figure 3 would fit right in with the first lady's branding aesthetic, which includes a self-titled coffee table book and movie, not least because the name "MELANIA" was emblazoned on the side of its glossy plastic head. After Figure 3 teetered gingerly away, Mrs. Trump looked around the room and told them that the future looked a lot like what they had just witnessed. 'The future of A.I. is personified,' she told her audience. 'It will be formed in the shape of humans. Very soon artificial intelligence will move from our mobile phones to humanoids that deliver utility.' She invited her guests to envision a future in which a robot philosopher educated children."

Read more of this story at Slashdot.

CNBC reports: Amazon has acquired Rivr, a Swiss robotics company developing machines for "doorstep delivery," the company confirmed Thursday... It announced the deal in a notice sent to third-party delivery contractors... "We believe this technology, when working alongside your [delivery associates], has the potential to further improve safety outcomes and the overall customer experience, particularly in the last steps of the delivery process...." In its notice to delivery service partner owners, Amazon said Rivr's technology, which includes a four-legged robot on wheels, will allow it to research and test how the devices can be integrated into delivery operations, including "helping [delivery associates] carry packages from delivery vehicles to customer doorsteps."

Read more of this story at Slashdot.

Des incidents récents survenus chez un développeur indépendant et, plus largement, au sein de grandes entreprises comme Amazon, relancent les questions sur l’usage de l’IA dans le développement logiciel.

Dix ans après avoir sécurisé WhatsApp, le créateur du protocole Signal, Moxie Marlinspike, récidive. Son nouveau projet, Confer, va s'intégrer à Meta AI pour garantir que vos échanges avec l'intelligence artificielle restent strictement privés, même pour Mark Zuckerberg.