Articles

SpeciesNet, Google's open-source AI model, identifies ~2,500 species in camera trap photos — used by conservation groups worldwide from Serengeti to Colombia to process millions of images.

Gemini in Google Sheets achieves 70.48% on SpreadsheetBench — state-of-the-art for AI spreadsheet manipulation, nearing human expert ability.

Google invests $1M AUD in AI-powered Population Health AI to identify hidden heart disease risks in rural Australia, where remote communities face 60% higher mortality rates.

Google DeepMind launches Nano Banana 2, combining Pro-quality image generation with Flash speed — featuring subject consistency, text rendering, and real-time world knowledge.

Gemini 3.1 Flash-Lite delivers 2.5X faster first-token speed than 2.5 Flash at $0.25/1M input tokens, with state-of-the-art benchmark scores for its tier.

arch 10, 2026 Research Demis Hassabis Ten years ago, our AI system AlphaGo became the first program to defeat a world champion at the complex game of Go - reaching a milestone in the field a decade be

OpenAI reframes prompt injection defense as social engineering risk management — designing systems where manipulation impact is constrained architecturally, not just filtered at input.

Rakuten achieves ~50% faster incident recovery using OpenAI Codex, integrating it across CI/CD, monitoring, and full-stack development workflows.

For decades, static application security testing (...

Practical lessons from treating Google AI Studio as a coding teammate — why setting boundaries with AI tools matters more than maximizing their output.

AMD runs the 1T-parameter Kimi K2.5 model locally across 4 Ryzen AI Max+ nodes (480GB total VRAM) using llama.cpp RPC — no cloud required.

Simon Willison proposes "interactive explanations" — animated visualizations of AI-generated code — as a pattern for paying down cognitive debt from agentic coding.

This is a brief guide to my new art project [microgpt](https://gist.github.com/karpathy/8627fe009c40f57531cb18360106ce95), a single file of 200 lines of pure Pyt...

As AI companies engage with defense contracts, enterprises should evaluate vendor transparency, usage policies, and develop their own AI governance frameworks.

GYWI combines author knowledge graphs with RAG + GraphRAG to generate more novel and feasible scientific ideas, evaluated across 5 dimensions on arXiv data.

No summary available

Google adds post-quantum ML-DSA signatures to TLS certificate transparency logs using Merkle Tree Certificates, keeping sizes at ~4KB while protecting against future quantum attacks.

Google Opal introduces dynamic agent steps with adaptive routing, persistent memory, and human-in-the-loop orchestration — a reference architecture for enterprise AI agents.

Unsloth Dynamic v2.0 quantization outperforms leading methods on multiple benchmarks, enabling accurate quantized LLM inference on consumer hardware.

Patrick Kidger wrote a [post](https://kidger.site/thoughts/just-know-stuff/) I read as I was getting into JAX, and even though the post was largely unrelated to what I was looking for, the writing its