Anthropic has developed new defenses that cut distillation data harvested from Claude by 90 to 99 percent. The company identified nearly 200 million distillation-related exchanges across five separate campaigns. The largest, attributed to Alibaba’s Qwen, involved over 151 million exchanges between May and July 2026. These defenses include strengthened extraction classifiers, summarization features, and a Fable 5.1 innovation called preserved thinking that prevents tampering with encrypted reasoning context. OpenAI faces the same challenge and released a September 2026 report outlining similar techniques, while questions remain about whether safety guardrails transfer to distilled copies.
Source: Read the original article
Disclaimer: this content is for information purposes only and is not financial advice. Cryptocurrencies are highly volatile: you may lose all of your capital. Always do your own research. Legal notice
