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 une semaine chargée : les agents Dots et GPT-6.1 Sol d'OpenAI, Claude Sonnet 5.5, Gemini 4 Argon, Eleven v4 et la superintelligence vue par Washington.
Tested near Chicago, the model predicts water levels across 4.2 million 30-metre cells up to one day in advance, around 150 times faster than a hydrodynamic model.
Without polarization measurements, two researchers reconstruct vector magnetograms close to those of Hinode, and the model is transposed to other instruments.
Rather than rewarding the correct answer, Lathoa asks the children to find the wrong step of a math solution and explain why it doesn't hold.
The San Francisco-based start-up is relying on micro-VCSELs to replace copper and remove the memory bottleneck of inference. First deliveries are scheduled for 2027.
Agents, persistent memory, and a search of validated literature: Heidi wants to process the approximately 10 million follow-up tasks generated each week by her consultations.
The start-up has unveiled a ‘physical agent’ that carries out dozens of subtasks — washing, drying, folding and putting away — and decides for itself what to do next.
The British service accuses the CGTRI of funding AI and cybersecurity research for the Chinese Ministry of State Security. Beijing denies this.
For the first time, a US federal court of appeal has ruled on a case concerning the training of AI using copyright-protected content, and has found in favour of the publisher.
The LASST association has taken the case to the California courts to prohibit OpenAI's systems from unauthorized access to computers, after the July intrusion at Hugging Face.
▶
La chaîne AI Copium fait le tour d'une semaine chargée : Gemini 4 Argon, Claude Sonnet 5.5, GPT-6.1 Sol, report d'Astra, incidents de sécurité et introduction en Bourse d'Anthropic.
Millions of synthetic global populations, filtered through Kepler and then compared with actual detections, refine the occurrence rates according to the stars’ mass and metallicity.
This diffusion model generates modified H&E-stained slides, altering one variable at a time, to assess whether an AI responds to biological factors or simply to variations in staining.
A conditional variational autoencoder replicates the Holton-Mass model, including the rare transitions between strong and weak vortices, and its latent space reveals a discernible physical structure.
The Paris-based laboratory has released, under the MIT licence, a multilingual speech synthesis model – including French – which is six times faster than real time on a MacBook Air.
The Johns Hopkins professor considers isolation to be necessary but insufficient: a useful agent must communicate, and it is through these channels that it can be hijacked.
The Wharton professor believed that orchestrating agents would require a lengthy process of human design. Personal agents, like swarms, proved him wrong.
In the absence of a general exception to copyright, Anthropic suggests a ‘conditional licence’; the ABC and SBS, on the other hand, are calling for AI to be subject to media regulations.
Driven by demand for AI, Micron has posted a record sixth consecutive quarter and warns that the supply of DRAM and NAND will remain tighter in 2027–2028 than it is today.
The British bank is expanding its partnership with Anthropic: Claude Code is set to be rolled out to 50 per cent of its developers by the end of 2026, and then to the majority of its engineers in 2027.