A newsletter breaking down AI research, technology, and Australian property in plain English
New arXiv framework uses twelve reasoning angles to separate jokes from hate in memes, hitting 80.3% humor and 75.9% hate detection on two benchmarks.
GigaPath-Flash and GigaTIME-Flash shrink billion-parameter pathology models to run on 50x less compute while predicting tumor profiles from routine slides.
MOSAIC, posted on arXiv on July 21, lifts long-conversation accuracy 27 points and catches 66% of factual conflicts before they corrupt an agent's memory.
A neural network trained on 314,630 demonstrations controls 25 systems from underwater vehicles to chemical reactors, matching hand-tuned controllers in
NVIDIA's 102.4-terabit Spectrum-6 Ethernet switch is reaching gigascale AI factories first at CoreWeave, Microsoft and Nebius, doubling network capacity.
New arXiv preprint shows AI agents can be redirected by reordering true facts, with 83.3% success across GPT, Claude, Gemini, DeepSeek and Qwen.
A new arXiv preprint maps 43 operations that corrupt self-hosted AI agents via legitimate OS calls, finding a residual surface no defense can distinguish.
LaCache caches unchanged tokens during diffusion LLM denoising, cutting inference 1.3X standalone and up to 40X stacked with other acceleration methods.
A July 2026 arXiv paper proposes that human raters' emotional states leak into preference labels used to train AI, with a five-prediction audit plan.
