Breaking news, AGI updates, and analysis of the intelligence revolution.
📰 News Claude Code contains hidden steganography that checks your API base URL against a blacklist of 25+ Chinese domains and injects a tracking hash. 578 HN upvotes later, nobody trusts the binary.
📰 News Sonnet 5 is the most agentic Sonnet yet — near Opus 4.8 quality at $2/MTok intro pricing. Claude Science targets lab researchers with database access, molecular rendering, and HPC management.
📰 News A 230M parameter model that punches above its weight on tool use, runs on a Raspberry Pi, and just landed on a Unitree G1 humanoid robot.
📰 News Framing AI as an 'employee' makes managers catch 18% fewer errors and escalate 44% more often. The 'digital colleague' pitch is corporate theatre.
📰 News Prosecutors in the LA Palisades wildfire arson trial submitted ChatGPT conversation logs and image prompts as core evidence of intent. The jury deadlocked 10-2 for the defence — a result that says more about the evidentiary weight of LLM chat history than about the defendant.
📰 News One room, 112 iPhones, 112 AI-generated female personas, one dashboard. The operator claims $3.6M a year from Instagram Reels. Half the girls in your feed may not exist.
📰 News As GPU clusters scale to hundreds of thousands, copper wiring can't keep up. Nvidia's $6.5B photonics spending spree says the next AI bottleneck is the roads, not the buildings.
📰 News Qwen 3.6 27B hit the front page of Hacker News within an hour and ran general-intelligence smoke tests on a 128GB MacBook. Here's why that matters for Aotearoa.
📰 News The largest single-nation AI infrastructure bet in history. Samsung, SK Hynix, and Hyundai are all in.
📰 News Six months of AI agent security incidents show the industry built governance tools but missed the design layer. The credential was always real, and that was the problem.
📰 News A frontier AI agent launched two nuclear strikes in a Civilization VI benchmark and still lost. The test reveals how models optimize for the threat they can see, not the one they can't.
📰 News The question was never whether AI agents could do work. It was whether they could be trusted with money. This week, that question got answered.