Anthropic cuts Claude token theft 90-99%
Trending storyDeveloping
Anthropic cuts Claude token theft 90-99%
1 reports1 sourcesUpdated 2 hours ago
What you need to know
AI summary
Anthropic says new defenses cut the volume of reasoning tokens extracted from its Claude models by 90 to 99 percent, according to The Decoder. The measures target industrial-scale distillation, in which third parties harvest a model's reasoning output to train competing systems. The figures are Anthropic's own claim; the report does not specify which defenses were deployed or how the reduction was measured.
Generated by AI from the reporting · updated 1 hour ago
Timeline
Follow the reports to see every side of the story.
Oct 13
- The DecoderAnthropic says it cut Claude reasoning token theft by up to 99% in fight against industrial-scale distillation
Anthropic says its new defenses cut the volume of extracted reasoning tokens from its Claude models by 90 to 99 percent.
Heat trend for this story
Not enough continuous data yet to chart a trend.