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
▶
Matt Wolfe passe en revue les actualités IA qui agitent le secteur cette semaine : annonces de modèles, produits et controverses, remises en perspective.
▶
Un résumé quotidien des sept actualités IA les plus marquantes du 15 août 2026, dans un format court et rythmé.
Since August 14, Pro, Max and Team subscribers have started their sessions with the agent authorized to act unless an irreversible or destructive action is detected.
The Identity Theft Resource Center counts 1,803 breaches in the first half of 2026 and 471 million notifications to victims. IBM estimates that a quarter of leaks are facilitated by AI.
Z.ai wanted a better model for identifying isolated faults; he obtained a system that links the stages of a complete operation. The oldest fault found dates back to 1981.
▶
Format court : le point quotidien sur l'actualité de l'IA, avec en fil rouge le mouvement discret des entreprises qui délaissent les modèles les plus chers.
▶
Tom Mackenzie enquête sur la robotique européenne : humanoïdes, machines autonomes et micro-robots chirurgicaux, entre laboratoire et déploiement industriel.
Four CVEs of 8.3 to 9.9 on Pyodide sandboxes, ten vulnerabilities in llama.cpp, and a Zoom exploit chain built in less than 24 hours with public models.
Google's standalone agent, previously reserved for the AI Ultra plan, becomes available to American AI Pro subscribers: it works in the background on Workspace services.
This is the first decline in three years in this market, while investments in AI increased the group's emissions by 25% last year.
Open since 2004, this numerical linear algebra conjecture gave way to a neurosurgery resident with no training in advanced mathematics, guiding a model in autonomous operation.
A model developed at NJIT reads the acoustic waves and magnetic field captured by NASA's solar observatory to anticipate the emergence of areas that cause solar storms.
Soundtrack, Music, SFX and Speech cover video soundtrack, song, sound effects and synthetic voice, with generation times much lower than the time listened to.
Hugging Face's half-year report shows 2.96 million model filings, 151,448 Qwen derivatives, and a striking size gap between Chinese and U.S. publications.
The Chinese laboratory publishes an update obtained without retraining the basic model, with 743 billion parameters and a context of one million tokens.
The Mobius-v0 architecture stores knowledge in a shared memory and entrusts reasoning to iterative modules, with a gain of about four times in inference speed.
Funded by the National Endowment for the Arts, the project aims to establish benchmarks for art schools confronted with the generalization of generative models.
After the teachers in the spring, students on all four campuses have had access to them since August 14, 2026. Prior training is mandatory, and the choice of allowing the tool remains with each teacher.
By tracking slope breaks in several time series, METR finds a spectacular acceleration in cybersecurity, plausible in mathematics, and unobtainable in algorithmics.
A test bench tested on 14 models shows that an accuracy of more than 95% does not protect against the influence of a plausible figure slipped into the question.