Anthropic Reports Leak of Account Data for About 100 Korean Claude Business Customers

At a glance

US AI company Anthropic reported to Korean regulators that account information for about 100 Korean business customers using Claude was exposed. The Personal Information Protection Commission received the report and started a preliminary review.

Key points

1. 1

   The exposed data includes business contacts' names and emails, account and organization IDs, plan tiers, and usage and spending figures.

2. 2

   Anthropic says chat contents, account passwords, and payment card and bank details were not included.

3. 3

   The leak started in external work software Anthropic uses, not in the Claude service itself.

4. 4

   Anthropic blocked the vendor's access, fixed the weakness, and received written confirmation that the data was deleted.

Why it matters

AI services that rely on many outside tools can leak customer data if just one connected tool is breached, which matters to Korean business customers.