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
A scene programme may run correctly whilst producing an impossible layout. VIGOR checks for collisions, clearances and structural relationships after execution.
A world model helps the agent choose which leads to execute. The actual results are then used to improve prediction and scientific exploration.
A defense without re-training weakens malicious memories while consolidating useful ones. The results cover back door attacks and injections.
Instead of building its response only from left to right, the agent refines a complete trajectory. It is evaluated on six chemistry tasks with external tools.
A specialist auditor analyses the actions of a code agent. In the test selected, classification accuracy rose from 57 per cent to 78 per cent following training.
Personalised tutoring establishes a clear link between observation, interpretation and teaching. RegulaTutor aims to provide long-term support rather than simply storing chat history.
A study of 750 million simulated collisions links latent anomalies to physical observables. The challenge is to make the warnings usable by physicists.
Ten combinatorics problems are used to improve verified Lean bounds. A solution becomes a new record to be broken instead of removing the problem.
Two local models propose and criticize scenarios and refine them from simulations. The approach aims at traceable test evidence on a laptop.
PRISM combines measurements, ECG data and population-based data. The study distinguishes its best scores from those obtained using the same input window as the systems under comparison.
An open framework combines independent peer review, a specialised wiki and Lean verification. The authors report high scores on several scientific tests from 2026.
Two resources combine Ukrainian images and text to evaluate persuasion. The study also compares the explanations of the models to the indices they mobilise.
The PAGE frame links automatic reading of the plates and human verification. Three studies cover 3,929 American comics and 87 volumes of Japanese manga.
An agent can fulfill a legitimate request while causing damage elsewhere. X-AgentRed explores the tool chains and tests them by execution.
▶
Un serveur MCP avec un outil par endpoint a cédé dès que toute l'équipe s'en est servie. Abhi Arya raconte comment Reducto l'a reconstruit autour de moins d'outils et d'une incertitude affichée.
On Truth Social, the American president hardened his terminology campaign: the White House would consider as adversaries those who stick to the consecrated expression.
Researchers compare eleven pre-trained interatomic potentials to DFT calculations and show that their errors explode when the atoms are compressed.
According to Crunchbase, AI captured $18.8 billion of the funds raised in Europe, a record share, driven by Mistral AI, Nscale and Helsing.
Remote-controlled demonstrations, low success rates, sceptical experts: Jamie Condliffe puts the promises of a humanoid robot in every home into perspective.
The new usage policy targets users who mistreat Claude for no apparent reason, and includes a section dedicated to safeguarding democratic processes.