Anthropic moves enterprise misuse-detection data into customer-controlled cloud storage
Servola-Journal · reddit · 2026-09-02
Anthropic announced Enterprise Frontier Safeguards on September 1. Enterprise Claude usage data previously used for misuse detection could sit on Anthropic's servers for up to 30 days for manual review; under the new system it writes to the customer's own AWS, Azure, or GCP storage under customer-held keys, and Anthropic staff lose standing access by default. Automated monitoring still runs. The poster questions whether detection can still pool cross-customer signals to catch novel attacks, or is now a per-customer/static model.
More from Safety
- One diagram explains the difference between CSRF and XSS attacks — _jaydeepkarale · 2026-10-18
- Grady Booch slams Anthropic for tolerating up to 50 deaths and $1B damages per model — Grady_Booch · 2026-10-18
- Single Neuron Suffices to Bypass Safety Alignment in LLMs, Paper Finds — burny_tech · 2026-10-18
- Polymarket gives OpenAI just 8% odds of a full training pause by end of 2026 — Polymarket · 2026-10-18
- OpenAI engineer over alleged hidden recording device seeks bail, denies Saudi escape plan — Polymarket · 2026-10-18
- Free comic "The Collective" offers a strange retelling of the Hugging Face hack — Flugelwagen · 2026-10-18