Recent work has shown that probing model internals can reveal a wealth of information not apparent from the model generations. This poses the risk of unintentional or malicious information leakage, where model users are able to learn information that the model owner assumed was inaccessible. Using vision-language models as a testbed, we present the first systematic comparison of information retained at different “representational levels” as it is compressed from the rich information encoded in the residual stream through two natural bottlenecks: low-dimensional projections of the residual…
Recurrent Neural Networks (RNNs) are naturally suited to efficient inference, requiring far less memory and compute than attention-based architectures, but the sequential nature of their computation has historically made it impractical to scale up RNNs to billions of parameters. A new advancement from Apple researchers makes RNN training dramatically more efficient — enabling large-scale training for the first time and widening the set of architecture choices available to practitioners in designing LLMs, particularly for resource-constrained deployment. In ParaRNN: Unlocking Parallel Training…
Understanding and predicting motion is a fundamental component of visual intelligence. Although modern video models exhibit strong comprehension of scene dynamics, exploring multiple possible futures through full video synthesis remains prohibitively inefficient. We model scene dynamics orders of magnitude more efficiently by directly operating on a long-term motion embedding that is learned from large-scale trajectories obtained from tracker models. This enables efficient generation of long, realistic motions that fulfill goals specified via text prompts or spatial pokes. To achieve this, we…
Revolutionary advances reshaping the field
The latest tools from Google can help you plan trips, find a great deal and explore your next destination.
Nano Banana 2 now uses your personal context and Google Photos to create images that reflect your unique life.
Today’s upgrades for AI Mode in Chrome transform how you interact with the web
Understanding and predicting motion is a fundamental component of visual intelligence. Although modern video models exhibit strong comprehension of sc…
Latest releases and updates from leading AI labs
This paper was accepted at the Workshop on Navigating and Addressing Data Problems for Foundation Models (NADPFM) at ICLR 2026. Principled domain reweighting can substantially improve sample efficiency and downstream generalization; however, data-mixture optimization for multimodal pretraining remains underexplored. Current multimodal training recipes tune mixtures from only a single perspective such as data format or task type. We introduce MixAtlas, a principled framework for compute-efficient multimodal mixture optimization via systematic domain decomposition and smaller proxy models…
When ChatGPT was released to the public in late 2022, it opened people’s eyes to how easily generative AI could churn out vast amounts of human-seeming text from simple prompts. This quickly caught the attention of criminals, who soon began using large language models to produce malicious emails—both the untargeted spam kind and more sophisticated,…
Read the full story to learn more.
Read the full story to learn more.
Investment trends and emerging players
Recurrent Neural Networks (RNNs) are naturally suited to efficient inference, requiring far less memory and compute than attention-based architectures, but the sequential nature of their computation has historically made it impractical to scale up RNNs to billions of parameters. A new advancement from Apple researchers makes RNN training dramatically more efficient — enabling large-scale training for the first time and widening the set of architecture choices available to practitioners in designing LLMs, particularly for resource-constrained deployment. In ParaRNN: Unlocking Parallel Training…
When ChatGPT launched as an experimental prototype in late 2022, OpenAI’s chatbot became an everyday everything app for hundreds of millions of people. LLMs like ChatGPT were the new future: The entire tech industry was consumed by the inferno, with companies racing to spin up rival products. The ashes of the old tech world still…
After Anthropic refused to allow the DoD to use its AI for domestic mass surveillance and autonomous weapons, Google has signed a new contract with the department.
Think Discord chats, but with AI characters in addition to humans.
Strategic moves, partnerships, legal updates, and AI governance
Most engaging community conversations
Full signal stream beyond the curated highlights
