Wells Fargo Ships Section 15 on November 24: Every AI Agent Instruction Is Deemed Authorized by You
On October 3, 2026, Wells Fargo posted a revised Online Access Agreement. Buried in it is Section 15: when an AI agent moves your money, you…
WMT26 by the Numbers: 46 Systems, 23 Directions, and the 29.30 BLEU Ceiling on Low-Resource Translation
What does machine translation actually deliver for a language your vendor does not care about? WMT26 put 46 systems through human annotators,…
Q3 2026 AI Funding by the Numbers: $103.3B, 85% in 154 Mega-Rounds, and Europe’s 75% Record
The Q3 2026 AI funding numbers do not describe one market. They describe two, and the figure that separates them is 85%.
Three Weeks Inside OpenAI’s Agent Monitor: What METR Broke, and the 10% Codex Gap Nobody Logged
OpenAI’s internal Codex traffic was 99.2% eligible for misalignment monitoring. Roughly 10% of it went unmonitored for three months…
AI Model Comparisons: 5 Myths, From the Benchmark OpenAI Retired to the $0.08 Model That Scores 21
On 23 February 2026, OpenAI explained why it had stopped reporting scores on the coding benchmark everyone quoted. Five months later it…
OpenAI Ships EU Text Watermarking, but Only Approved Researchers Get the Detector
OpenAI said on 27 October 2026 it will start watermarking AI-generated text in the EU. The detector goes to approved researchers and expert…
CrossDiT, Explained: How Kandinsky 6.0 Generates Video and 44 kHz Audio in One Pass, and Where It Breaks
Kandinsky Lab shipped a 29B audio-video model under MIT on 6 October 2026, weights and code included. The clips it produces are five seconds long.
IBM Defense Model in Practice: What the Model Card Says, and What the Pricing Page Does Not
Janes says the model cuts research time “up to 85%”, with no methodology, baseline, dataset or independent validation attached.…
Setting up a self-hosted ComfyUI image pipeline: a CTO’s guide to licences, GPU costs and node risk
What exactly do you own when you self-host ComfyUI? The core is GPL-3.0, and that licence covers the software and almost nothing else in your…
FrontierMath Erdős by the Numbers: 68 Open Problems, 2 Solved, 98% on the Benchmark Next Door
Epoch AI calls FrontierMath Tier 4 “approaching saturation” for GPT-6 Astra at 98%. The same model, handed 68 problems nobody had…
