Get the Brief
AnthropicIncidentMateriality 3

Anthropic cuts Claude reasoning token theft by up to 99%

Anthropic says it has reduced extraction of Claude's internal reasoning tokens by 90 to 99 percent. The company links large-scale distillation attacks to Chinese firms Zhipu and Moonshot and has banned tens of thousands of accounts.

First seen 13 Oct, 11:06 UTC on The Decoder2 sourcesLast update 4h ago

What the sources say

Linked, never rewritten. Official means the lab itself.

Official0

No word from Anthropic yet.

Press2

The Decoder· 13 Oct, 11:06Anthropic says it cut Claude reasoning token theft by up to 99% in fight against industrial-scale distillationThe Decoder· 15 Oct, 15:27Bank of America puts 4,000 people on fixing vulnerabilities uncovered by Anthropic's Mythos and other AI models

How it unfolded

Every source in the order it appeared. Times in UTC.

13 Oct 11:06The Decoder Anthropic says it cut Claude reasoning token theft by up to 99% in fight against industrial-scale distillation15 Oct 15:27The Decoder Bank of America puts 4,000 people on fixing vulnerabilities uncovered by Anthropic's Mythos and other AI models

What the numbers did

Signals for Anthropic from two hours before this story to ten days after: leaderboards, app charts, usage, prices and status.

StatusAnthropic's status page is back to normal

Keep up with Anthropic