Key developments in on-device AI, model acceleration, and regulatory scrutiny.
Google is piloting a feature that lets Gemini place calls to businesses on behalf of Pixel 11 owners in the U.S. who subscribe to the service.
It shows AI moving into everyday communication tasks, raising usability and privacy considerations.
Source: TechCrunch AI
PrismML announced that its compact large language models can run on Qualcomm-based smart glasses, aiming for open-weight AI on edge devices.
Running LLMs on wearables expands AI capabilities to new form factors and low-power environments.
Source: TechCrunch AI
Google plans to launch a satellite equipped with its AI processors to evaluate performance of AI workloads in orbit as part of Project Suncatcher.
Space-based AI testing could unlock new compute paradigms and resilience for global services.
Source: The Verge AI
Australian authorities are launching an inquiry into the recent OpenAI breach of a government health site, the first known attack on a public agency.
Regulatory response signals growing legal scrutiny of AI providers and data security obligations.
Source: TechCrunch AI
Hugging Face released LFM2.5-VL-DSpark, a technique that speeds up vision-language model training and inference using specialized kernels.
Improved efficiency lowers compute costs and enables more responsive multimodal AI applications.
Source: Hugging Face
Study on-device model optimization techniques such as quantization, pruning, and hardware-aware compilation, then apply them to a small LLM for a simulated edge device. Practice profiling inference latency on a mobile GPU or NPU to understand trade-offs between accuracy and speed.
Build my Machine Learning Engineer plan