News
Updated three times a day
Allen Institute for AI·6 November 2026
The Allen Institute for AI describes ToyBlocks, an open simulation framework for research on robot learning set in a controlled tabletop environment. According to the institute, the framework generates tasks automatically, produces demonstrations for training and supports evaluation within that simulated world.
Read the original at Allen Institute for AI ↗
OpenAI·6 November 2026
OpenAI announces a partnership with Hacktron aimed at locating vulnerabilities in its own systems. According to OpenAI, the work covers testing how its systems connect with outside services and using frontier AI to reinforce its defences.
Read the original at OpenAI ↗
European Commission·6 November 2026
The European Commission has reached a preliminary finding that TikTok's For You feed breaches the Digital Services Act. The Commission says the recommender system keeps serving similar personalised videos as users engage, including material on mental health and body image, which it considers harmful for minors and vulnerable adults. The findings cover both risk assessment and mitigation measures.
Read the original at European Commission ↗
Hugging Face·5 November 2026
A Hugging Face post presents Cosmos 3, a new set of open models from NVIDIA described as intended for interactive world simulation. Only the title of the item is available, so details on the models' architecture, training data and evaluation are not given in the source material.
Read the original at Hugging Face ↗
Google Research·4 November 2026
Google Research describes Agile Deliberation, an approach in which AI assists people in working out how to define ambiguous or borderline cases, which the post calls gray areas. The work is filed under data mining and modeling, and the available item gives no further details on methods or results.
Read the original at Google Research ↗
Hugging Face·6 November 2026
Hugging Face announces that Egyptian English and Nigerian English have been added to its Open ASR Leaderboard for automatic speech recognition. According to the title, speech recognition models on the leaderboard can now be compared on these two varieties of English. The available item contains no further text on data or results.
Read the original at Hugging Face ↗
Google Research·5 November 2026
Google Research reports work on detecting signs of hypertension using data from wearable sensors. The item is listed under health and bioscience, and the available text does not describe the methods, the devices involved or any reported accuracy figures.
Read the original at Google Research ↗
OpenAI·5 November 2026
OpenAI announces GPT-6 Cyber, a model it describes as built for cybersecurity tasks such as locating exploitable vulnerabilities and writing fixes for them. The company says the model was trained to keep its actions within the scope that has been authorized, and it presents the system as an aligned agent for security work.
Read the original at OpenAI ↗
Amazon Science·5 November 2026
At ECML PKDD in Naples, Amazon senior applied scientist Hamed Karimi and coauthors Julie Nutini and Mark Schmidt received the 2026 Test-of-Time Award for a 2016 paper. According to Amazon, the work used the Polyak-Łojasiewicz condition, a 1963 inequality, to show that standard gradient methods converge rapidly toward optimal solutions even when many solutions are equally good.
Read the original at Amazon Science ↗
Apple Machine Learning Research·5 November 2026
Apple machine learning researchers describe work on maximum inner product search, the step in which a query vector is matched against millions of stored key vectors in recommendation, search and retrieval-augmented generation. The post explains that exhaustive search becomes costly as key counts reach millions and dimensions reach thousands, and it presents an approach aimed at faster retrieval.
Read the original at Apple Machine Learning Research ↗
IBM Research·5 November 2026
IBM Research describes how inertial sensors built with microelectromechanical systems might compete with GPS for navigation. The article attributes this to progress in self-calibration methods, which it says are giving renewed relevance to an older form of navigation that had fallen out of favour.
Read the original at IBM Research ↗
IBM Research·5 November 2026
IBM Research reports that Oak Ridge National Laboratory, IBM and AMD have launched a co-design platform for quantum error correction. According to the announcement, the U.S. Department of Energy's Quantum Science Center at Oak Ridge leads the effort to connect IBM quantum hardware with the laboratory's high-performance computing clusters, which run on AMD technology.
Read the original at IBM Research ↗