Anthropic Enterprise Frontier Safeguards pairs zero data retention with automated detection of remote misuse and keeps logs in customer-controlled cloud, according to the Sep 1 announcement.
It will roll out in phases this fall across Claude Code, Claude Enterprise, Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google Agent Platform and Microsoft Foundry, with ZDR on Fable 5 and Fable 5.1 as interim cover.
How the attack actually works
The hardest misuse doesn't happen in one prompt. It's spread across many sessions and accounts, so single-turn checks that discard data instantly miss the pattern.
Anthropic points to Mythos-class models like Fable 5.1, where agents can autonomously pursue destructive actions, and to cases of stolen enterprise credentials that only stand out when traffic is correlated.
Think of it like card fraud detection that ignores one swipe but flags a ring of small charges across stolen cards over weeks. That's why Anthropic moved to 30-day retention with Fable 5 before building EFS.
