A newsletter breaking down AI research, technology, and Australian property in plain English
Researchers show a 5G NR side-channel attack can halve a user's video stream quality by reading public scheduling data, without touching encrypted traffic.
A new preprint claims PagedWeight dynamically quantizes MoE model weights at runtime, saving 72% GPU memory and lifting throughput 1.94× for AI inference.
A July 2026 arXiv preprint proposes learning interpretable trustworthiness levels and monitoring them across an AI system's entire lifecycle.
A new arXiv study of OECD-catalogued AI trust tools finds the field heavy on fairness and transparency but thin on security, explainability and
NVIDIA's SIGGRAPH keynote unveils MCP connections letting AI agents work inside creative apps, plus Cosmos 3 Edge for local physical AI.
The Australia Institute asks why wages lag productivity. The answer reaches every mortgage application and rent payment in the country.
A new arXiv preprint claims to solve a hard problem in probability, running on a desktop CPU without neural networks. Scientists and engineers take note.
A new arXiv preprint introduces a method to measure whether frontier AI helps non-experts plan CBRN attacks, with uplift limited to radiological threats.
An arXiv preprint analysing 14 years of Qubes Security Bulletins finds most flaws originate in Xen and CPU components, not Qubes itself.
