Anthropic’s new safeguards cut AI distillation attacks on Claude models by 90-99%

Anthropic , the developer of the Claude AI models, announced that its new defensive measures have significantly reduced the impact of AI distillation attacks. According to a report by Reuters, the safeguards have made it substantially more difficult to extract reasoning tokens from Claude models, with improvements reducing data harvesting by as much as 90% to 99%. The recent measures are implemented in the Claude 5-series models, including Fable 5.1, Opus 5.5, and Sonnet 5.5, and have resulted in the banning of tens of thousands of accounts linked to attack campaigns.

BeCrypto News shows a short excerpt and links to the publisher. The full report belongs to Crypto Briefing.