Anthropic Accuses Alibaba of Distilling Claude to Train Qwen
2 videos · score: 2,970 · first seen Jun 29, 2026
Anthropic accuses Alibaba of scraping roughly 29 million Claude conversation exchanges to train its Qwen models via a distillation attack, raising IP and safety concerns about frontier-model outputs being harvested by competitors.
Fable 5 vs GPT 5.6 Sol: The Early Results
OpenAI's limited government-only release of GPT-5.6 Sol and Anthropic's re-release of Fable 5 with a stricter safety classifier, alongside Anthropic's accusation that Alibaba scraped 29 million Claude exchanges for Qwen training, highlights how distillation attacks may incentivize labs to delay general frontier model releases to prevent IP theft.
Claude got attacked
Anthropic's letter to the Trump administration reveals Alibaba's distillation attack on Claude, arguing it justifies pressuring OpenAI and Anthropic to slow new model releases.