AI;DR: AI news, ranked every 30 minutes
03AI;DRTUESDAY · NOVEMBER 2026
- anthropicAnthropic’s amended filing reports Q3 revenue of $16.96B, while NYT sources put it at $16.8B and say growth flattened. The company swung to a $567M operating loss as compute costs surged.+1
- anthropicGoldman Sachs is restricting Claude Code and GitHub Copilot to engineers, removing them from traders and bankers. The bank cited productivity and risk management after previously praising nontechnical “vibe coding.”
- regulationChina’s MOFCOM and MOST proposed export controls covering 23 AI technologies, including frontier model weights and agent architectures, while an interim registration rule already applies. Comments are open until November 15, 2026.
- llmAstra can natively read and write neural-network weights, performing beyond what researchers expected from humans or conventional algorithms. The capability was demonstrated while improving Qwen.
- fundingOracle plans to raise $20B through $10B of stock and $10B of bonds to build AI data centers for customers including OpenAI and repay debt. Its shares fell over 6%, trading about 54% below the yearly high.
- nvidiaNvidia told cloud providers leasing its hardware to sell at least 51% of services to customers outside China. It also urged data-center clients to reduce China exposure amid intensifying US-China AI competition.
- agentLongCat-2.5 is a 1.6-trillion-parameter model with native multimodality for agentic tasks. Its training was completed entirely on domestically made chips, highlighting China’s push for local AI infrastructure.
- safetyThe UK’s National Cyber Security Centre warned that machine-learning-based attacks on critical infrastructure could occur within 24 months. It said China is likely the world’s most capable actor for such operations.
220 storiesUpdated 12m ago