Microsoft AI researchers mistakenly leak 38TB of data

Published , by Donovan Erskine

Microsoft has been a major player in AI ever since it started to gain traction in the industry. The company has invested millions of dollars into artificial intelligence, and has a team dedicated to researching and making advancements in AI tech. Their operations led to an unfortunate blunder this week when Microsoft AI research accidentally leaked a whopping 38TB of company data.

The massive data leak from Microsoft was first spotted by Wiz on Github, and Microsoft posted an official statement shortly thereafter. The leak apparently happened inadvertently while researchers were trying to upload open-source code that would help other researchers train AI models. Among the information leaked was “passwords to Microsoft services, secret keys, and over 30,000 internal Microsoft Teams messages from 359 Microsoft employees.” In Microsoft’s statement, the company assures customers that the leaks did not include any of their sensitive information.


Source: Getty Images

Microsoft also explained how a Shared Access Signature (SAS) token lead to the leak and how this situation will help them avoid similar leaks in the future.

In an unfortunate twist of irony, this isn’t even the only notable Microsoft leak in the span of 24 hours. Last night, a massive Xbox leak revealed documents and emails pertaining to the inner workings at Microsoft’s gaming division. This included images of an upcoming refresh of the Xbox Series X, several unannounced games from Bethesda Softworks, and an email in which Phil Spencer expressed interest in acquiring Nintendo.