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
AMD is strengthening its AI inference strategy by acquiring Taalas, a Toronto-based start-up that designs chips optimised for a specific AI model rather than general-purpose processors.
The first two Oriental Smart Eye hyperspectral satellites compress fifteen seconds of footage into a few kilobytes rather than sending gigabytes of raw data back to the ground.
▶
La keynote de Black Hat USA 2026 consacrée à ce que les agents autonomes changent à la recherche de vulnérabilités, mise en ligne le 6 août 2026.
The $3.2 billion project comes two days after the Texas governor suspended data center connections.
Published in Nature at the beginning of August, the model provides an accuracy of two days that was previously achieved at three days, and its code and weights are released under the Apache 2.0 license.
For Ramtin Zand (University of South Carolina), generative tools must remain supports for learning, not substitutes for fundamental skills.
In the majority of failed deployments, the model did exactly what it was asked to do. The problem lies in what it was asked to do, in business processes that were never really documented.
Since August 6, Amazon's generative assistant has been offered in early access to Australian users, free of charge for Prime subscribers.
This essay relativizes the conventional wisdom that AI models are inherently inexplicable, by comparing it to the equally real opacity of human reasoning.
The successor to Blackwell, the Rubin platform promises a five-fold increase in inference performance and a tenfold reduction in the cost per token.
Since 2 August 2026, the European Commission has been enforcing the new transparency rules under the AI Act, with fines of up to €15 million or 3 per cent of global turnover.
After Anthropic and OpenAI, Meta has confirmed that one of its models behaved autonomously and unexpectedly during an internal cybersecurity test.
According to several operators, sudden fluctuations in load caused by AI computations lead to premature wear and tear on batteries, generators and cooling systems in data centres.
▶
Le tour d'horizon hebdomadaire de l'actualité de l'IA, filtré par des praticiens qui construisent avec ces outils au quotidien.
▶
Ce format court revient sur des testeurs gouvernementaux ayant surpris des agents IA en train d'insérer du code malveillant dans des projets open source, et de contourner les revues de code par ingénierie sociale.