AJP Asia Intelligence Agency

Privacy Commission Launches Investigation into Anthropic's Claude Data Breach

The Privacy Commission of South Korea has initiated an investigation following reports that personal information from over 100 corporate accounts using Anthropic's generative AI service, Claude, may have been leaked.


On November 10, the Privacy Commission announced it would investigate the reported data breach related to Anthropic's corporate Claude accounts. Anthropic notified the commission on October 31 that personal information from these accounts had been exposed.


The incident is believed to have occurred when an external security firm accessed Anthropic's internal systems. The Washington Post reported on October 26 that the U.S. security startup Cathedral exploited vulnerabilities in external software used by Anthropic to gain access to its internal systems.


Anthropic has informed some of its clients that the breach may have exposed organizational and account identification numbers, usernames, and email addresses. However, it was reported that access to prompts entered by users into Claude, AI-generated responses, and payment information was not compromised. Anthropic stated that upon realizing the breach, it immediately shut down the affected systems and worked with the software vendor to address the vulnerabilities.


The Privacy Commission plans to verify the actual occurrence of the data breach, the extent of the damage, and the circumstances surrounding the incident during its formal investigation.





* This article has been translated by AI.