4 hours ago
- Ben Thompson proposes that the U.S. legalize data collection for training AI models as fair use and ban terms of service that prohibit distillation, to help US open models compete with Chinese counterparts.
- Distillation, which involves querying APIs, is nearly impossible to stop, so the U.S. should embrace a copyright policy that indemnifies labs and promotes further innovation.
- Alibaba's release of Qwen 3.8 Max as open weights may have been influenced by Xi Jinping's speech encouraging open source and collaboration.
- Qwen 3.8 Max is a 2.4T parameter model, nearly as large as Kimi K3's 2.8T, and its reasoning trace included humorous self-corrections like 'Could add helmet? No.'
- Recent articles mentioned cover topics like OpenAI's accidental attack on Hugging Face, a chat with the Claude Code team, and Kimi K3's pelican benchmark.