Nearly 250 million Microsoft Customer Service and Support (CSS) records were found exposed to the Internet in five insecure Elasticsearch databases, Comparitech reports.
The records on those servers contained 14 years’ worth of logs of conversations between support agents and customers, all of which could be accessed by anyone directly from a browser, without any form of authentication.
Each of the five Elasticsearch servers contained an apparently identical set of records, with data spanning between 2005 and December 2019, Comparitech’s security researchers reveal.
While most of the personal information in those records was redacted, many records contained plain text data.
Exposed data in those records included customer email addresses, IP addresses, locations, descriptions of CSS claims and cases, Microsoft support agent emails, internal notes marked as “confidential,” and case numbers, resolutions, and remarks, the researchers say.
“I immediately reported this to Microsoft and within 24 hours all servers were secured,” security researcher Bob Diachenko, who led the Comparitech team, explains.
The data was exposed to the Internet for around two days before Microsoft secured the servers. The databases were indexed by search engine BinaryEdge on December 28, 2019, Diachenko discovered them the next day and notified Microsoft, and the company secured the servers on December 30.
The exposed data could be abused in attacks such as tech support scams, where cybercriminals impersonate Microsoft support representatives. With this data in hand they could refer to real case numbers, phish for sensitive information, or hijack devices, Comparitech says.
“We’re thankful to Bob Diachenko for working closely with us so that we were able to quickly fix this misconfiguration, analyze data, and notify customers as appropriate,” commented Eric Doerr, general manager of Microsoft’s Security Response Center.
What is yet unclear is whether other unauthorized parties accessed the databases while they were exposed.
UPDATE. Microsoft says that the exposure was the result of a misconfiguration that occurred on December 5, but that its investigation into the incident did not reveal malicious use.
“Our investigation confirmed that the vast majority of records were cleared of personal information in accordance with our standard practices. In some scenarios, the data may have remained unredacted if it met specific conditions. […]We have begun notifications to customers whose data was present in this redacted database,” the company says.