Current developments in generative AI, software engineering, programming, and digital products, with links to original sources.
Researchers Demonstrate OpenAI Internal Repository Breach Using Claude Opus 5 — OpenAI Launches Law-Specific Astra, Google Overhauls Its Agent Infrastructure
SynthID Watermarking Alters LLM Safety Behavior — Jev Enters the Real-World Testing Phase, and the Key to Long-Running Agents Lies in the "Harness"
Amazon: "Release Models Only When They're Ready" — Palantir Argues "Personal Responsibility," Waymo Headed to Singapore in 2028
Claude Now Leads a Quarter of Next-Gen Model Development — 'Theft' Remark Surfaces in Microsoft Lawsuit as UK King Warns of AI Risks
OpenAI May Be Closing In on the Hodge Conjecture — GPT-5.6's “Misconduct-Hiding Memo” Disclosed, and Claude Code Moves to Parallel Agents
GPT-6 Astra Clears Pokémon in 18 Hours — AI Agents Make Phone Calls, and Huawei Moves Up Its New Chip
GPT-6 Astra Reportedly Decodes an Enigma Message Unsolved for 83 Years in 10 Hours — The “Coordination Tax” on Agent Sprawl, and Huawei's AI Chip Supply Squeeze
Tabular Foundation Model LimiX-2 Tops 3 Benchmarks — Local LLM Clinic Draws Attention, AI Agents Enter FPGA Design
Databricks Reveals Real-World Numbers from Its Astra Rollout — “Spend +60%”; a 29B Chinese Model Fully Trained on Ascend NPUs, and Cloudflare Blocking AI Training Only
Snap's $2,200 Smart Glasses Take Another Shot — US House Advances AI Power Cost Bill, Qwen 3.8 Seeks Out Browser Testing on Its Own
