AM-IA brings together news, research papers and applications relating to artificial intelligence published around the world into a single feed — and makes them available in your European language at the click of a button.
Scroll to browse · Click to filter
▶
Une ancienne investisseuse en dette en difficulté soumet le cas iHeartMedia à deux modèles de pointe : leurs pronostics divergent, jusqu'à ce qu'un raisonnement découpé en couches vérifiées les fasse converger.
A study of 27 employees from 21 SMEs shows that access to AI is not sufficient: verification, traceability and repair of local contraindications are a condition for the announced gains.
By placing up to 100,000 agents from four LLMs on different networks, physicists show that topology dictates the speed of consensus and vulnerability to manipulation.
A survey of 313 students, supplemented by 18 interviews, describes very uneven learning paths, and a skill that resists: knowing how to evaluate what AI produces.
Four researchers propose a common language, aligned with ISO/IEC standards, to catalogue AI uses and their risks, and to facilitate compliance with several regulations at once.
Out of 500 questions from 346 astronomy articles, five models reach nearly 90% of the right answers in easy versions, but cap around 60% as soon as the authors' explanation is only partial.
This method combines a learned world model, charged with anticipating mobile obstacles, with a differentiable quadrirotor simulator, and converges faster than existing approaches in mobile door racing.
The SIM-UT method simulates thousands of realistic exchanges with three chatbots deployed in India, South Africa and Kenya, and reveals a clinically dangerous convenience at MedGemma as well as ChatGPT and Claude.
A study shows that couriers who foil traditional phishings give way to persuasive, custom-designed emails, to the point of executing malicious demand.
The Parisian start-up launches two models of 31 and 40 million parameters, multilingual and licensed under Apache 2.0, designed to speak without connection on mobile, PC or browser.
▶
Chez Arca, société de services financiers « native IA », des agents prennent en charge des objectifs clients de bout en bout sur plusieurs jours et sollicitent les humains au bon moment.
The tool translates into scores readable by non-engineers twelve technical requirements of the AI Act measurable by benchmarks, and also measures the energy consumed by each assessment.
Researchers publish a 3D touch sensor and its entire software ecosystem; a CAD tool generates manufacturing files from a contour or text description.
By analyzing 429 requests from 23 first-year students, two researchers show that the copies rendered say little about the mathematical work actually delegated to the model.
Tested on 86 clinical tasks in 20 languages, models that directly attribute a probability to each answer are fast and well calibrated on the text, but pick up on the images and face the triage nurses.
Researchers have blindly read articles published in 2026 in major journals: models demand massive revisions, contradict each other and often miss errors that have been deliberately slipped.
Faced with Nilay Patel, Microsoft's co-founder believes that competition between laboratories makes self-regulation impossible and that only public authorities can impose safeguards.
Claroc researchers have reconstructed shared links of the Muse Assistant from public DNS data, outlining management notes, loan files and bank alerts. Meta suspended the function.
Vercel's October barometer shows AI budgets that explode despite falling prices, major open models in volume and one in two teams that changed their main model in September.
Block's IA lab presents Dessa, a full-duplex speech-to-word model learned without any text, able to listen and respond at the same time on a laptop.