Become a Member

Regulation TrackerLaws, orders and treaties read for their security duties and clocks Standards & FrameworksAttack taxonomies, control catalogues and frontier frameworks AI Incident DatabaseThe security incident record, with mechanism and lesson Research Paper LibraryThe papers that define the field, summarised OSS DirectoryRed teaming, guardrail, supply chain and evaluation tools Case StudiesProgrammes built, with measured outcomes AI Vulnerability TrackerCVEs and advisories for the components AI systems are built from

Latest in the tracker

New York Stop Addictive Feeds Exploitation (SAFE) for Kids Act and implementing rulesNew York's law restricting algorithmic feeds for minors takes effect on 25 January 2027 under final rules published in July 2026, with certified age assurance and $5,000-per-violation penalties.

Knowledge Base Regulation Tracker Standards & Frameworks OSS Directory AI Incident Database Research Paper Library Case Studies Society Articles News Jobs About Charter Become a Member

Home/AI Incident Database/Microsoft AI research team exposes 38 TB of internal data through a misconfigured Azure SAS token on GitHub

Microsoft AI research team exposes 38 TB of internal data through a misconfigured Azure SAS token on GitHub

A GitHub repository publishing open-source training data linked to an Azure storage token that granted full access to a 38 TB account, exposing workstation backups, passwords and thousands of internal messages for nearly three years.

Severity: Highmicrosoftsas tokentraining dataexposure

At a glance

Organisation: Microsoft
Industry: AI developers and research labs
Country or region: United States
Occurred: 20 Jul 2020
Discovered: 22 Jun 2023
Disclosed: 18 Sept 2023

All details 15

Type of harm: Confidential information and IP leakage, Credential and secret exposure, Privacy and personal data exposure
Kind of system: Training data and datasets
Severity: High
Confirmation: Confirmed by the organisation
Tools involved: Azure Blob Storage, SAS tokens
Technique: Over-permissive shared access signature on a training-data repository
Primary root cause: Misconfiguration or exposed service
Scale: 38 TB including two employees' workstation backups, 30,000 internal Teams messages and secrets
Outcome: Microsoft revoked the token, said no customer data was exposed and changed SAS token scanning; the case is a reference for data-sharing risk in AI research.

Sources

Contribute

Spot an error, or know a record we should add? Members can propose additions and submit corrections.

Submit incident Report a correction or request removal