


Up to 38TB of data accidentally leaked! Microsoft's AI research team 'poke trouble' How to ensure data security?
Financial Associated Press News on September 19 (Editor Xiaoxiang) Since the rise of generative artificial intelligence this year, there have been constant disputes about data security. According to the latest research from a cybersecurity company, Microsoft's artificial intelligence research team accidentally leaked a large amount of private data on the software development platform GitHub a few months ago, which included more than 30,000 pieces of internal information about the Microsoft team.
A team at cloud security company Wiz discovered that open source training data was leaked by Microsoft's research team when it was released on GitHub in June. The cloud-hosted data was leaked through a misconfigured link.
According to a blog post by Wiz, Microsoft's AI research team initially released the open source training data on GitHub. However, due to a misconfiguration of the SAS token, they incorrectly configured the permissions to grant the entire storage account and also gave the user full control instead of just read-only permissions. This means they can delete and overwrite existing files
According to Wiz, the 38TB of leaked data includes disk backups of the personal computers of two Microsoft employees. These backups contain passwords and keys for Microsoft services and more than 30,000 Microsoft team entries from 359 Microsoft employees. Internal information
Wiz researchers pointed out that open data sharing is an important part of AI training. However, large-scale data sharing can also pose significant risks to companies if not used correctly
Wiz CTO and co-founder Ami Luttwak pointed out that Wiz shared the situation with Microsoft in June, and Microsoft quickly deleted the exposed data. The Wiz research team discovered these data caches while scanning the Internet for misconfigured storage.
In response, a Microsoft spokesperson commented afterwards, "We have confirmed that no customer data has been exposed and no other internal services have been compromised."
In a blog post published on Monday, Microsoft said it has investigated and remediated an incident involving a Microsoft employee who shared a URL to an open source artificial intelligence learning model in a public GitHub repository. Microsoft said the exposed data in the storage account included backups of workstation configuration files of two former employees, as well as internal Microsoft Teams information about the two former employees and their colleagues.
The rewritten content is: Source: Financial Associated Press
The above is the detailed content of Up to 38TB of data accidentally leaked! Microsoft's AI research team 'poke trouble' How to ensure data security?. For more information, please follow other related articles on the PHP Chinese website!

Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

AI Hentai Generator
Generate AI Hentai for free.

Hot Article

Hot Tools

Notepad++7.3.1
Easy-to-use and free code editor

SublimeText3 Chinese version
Chinese version, very easy to use

Zend Studio 13.0.1
Powerful PHP integrated development environment

Dreamweaver CS6
Visual web development tools

SublimeText3 Mac version
God-level code editing software (SublimeText3)

Hot Topics



Vibe coding is reshaping the world of software development by letting us create applications using natural language instead of endless lines of code. Inspired by visionaries like Andrej Karpathy, this innovative approach lets dev

February 2025 has been yet another game-changing month for generative AI, bringing us some of the most anticipated model upgrades and groundbreaking new features. From xAI’s Grok 3 and Anthropic’s Claude 3.7 Sonnet, to OpenAI’s G

YOLO (You Only Look Once) has been a leading real-time object detection framework, with each iteration improving upon the previous versions. The latest version YOLO v12 introduces advancements that significantly enhance accuracy

The article reviews top AI art generators, discussing their features, suitability for creative projects, and value. It highlights Midjourney as the best value for professionals and recommends DALL-E 2 for high-quality, customizable art.

ChatGPT 4 is currently available and widely used, demonstrating significant improvements in understanding context and generating coherent responses compared to its predecessors like ChatGPT 3.5. Future developments may include more personalized interactions and real-time data processing capabilities, further enhancing its potential for various applications.

The article discusses AI models surpassing ChatGPT, like LaMDA, LLaMA, and Grok, highlighting their advantages in accuracy, understanding, and industry impact.(159 characters)

Mistral OCR: Revolutionizing Retrieval-Augmented Generation with Multimodal Document Understanding Retrieval-Augmented Generation (RAG) systems have significantly advanced AI capabilities, enabling access to vast data stores for more informed respons

The article compares top AI chatbots like ChatGPT, Gemini, and Claude, focusing on their unique features, customization options, and performance in natural language processing and reliability.
