


Microsoft AI researchers accidentally exposed a large amount of internal data due to misconfiguration of cloud storage links
On September 19th, the latest research from a cybersecurity company showed that Microsoft’s artificial intelligence research team accidentally exposed a large amount of private data on the software development platform GitHub. More than 30,000 internal messages communicated by Microsoft employees through Teams were leaked. Microsoft quickly deleted the exposure data after receiving the warning.
A team at cloud security company Wiz discovered that the data, which was hosted on the cloud platform, was exposed through a misconfigured link. According to Wiz, Microsoft's artificial intelligence research team inadvertently leaked this data when it released open source training content on GitHub.
According to reports, users of relevant repositories have been notified that they can download artificial intelligence models through the cloud storage URL link. However, according to a Wiz blog post, this link was misconfigured, resulting in the user having permissions for the entire storage account and also being granted full control over the entire repository, rather than just read-only permissions. This means users can delete and overwrite existing files at will. According to Wiz, the leaked data in the repository includes Microsoft employees' personal computer backup information, which contains Microsoft service passwords, keys, and more than 30,000 internal Microsoft Teams messages from 359 Microsoft employees
Wiz's Researchers say open data sharing is a key component of AI training efforts. But if used incorrectly, sharing large amounts of data can put companies at greater risk. Wiz CTO and co-founder Ami Luttwak said Wiz shared the news with Microsoft in June, and Microsoft acted quickly to remove the exposed data. He added that the incident "could have been much worse".
When asked to comment on the data breach, a Microsoft spokesperson said: "We have confirmed that no customer data was compromised and no other internal services were affected."
In a blog post published on Monday, Microsoft said that the incident involved a Microsoft employee sharing a URL from a GitHub public repository to an open source artificial intelligence learning model, and that the company has investigated and implemented remediation measures. . Microsoft said the data exposed in the storage account included backups of computer configuration files of two former employees, as well as internal Microsoft Teams messages between the two employees and their colleagues.
According to the blog, the data exposure was discovered by the Wiz research team while scanning the Internet for misconfigured storage as part of their efforts to target accidental exposure of cloud-hosted data. (Chenchen)
The above is the detailed content of Microsoft AI researchers accidentally exposed a large amount of internal data due to misconfiguration of cloud storage links. For more information, please follow other related articles on the PHP Chinese website!

Hot AI Tools

Undress AI Tool
Undress images for free

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

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

Clothoff.io
AI clothes remover

Video Face Swap
Swap faces in any video effortlessly with our completely free AI face swap tool!

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)

The core method of building social sharing functions in PHP is to dynamically generate sharing links that meet the requirements of each platform. 1. First get the current page or specified URL and article information; 2. Use urlencode to encode the parameters; 3. Splice and generate sharing links according to the protocols of each platform; 4. Display links on the front end for users to click and share; 5. Dynamically generate OG tags on the page to optimize sharing content display; 6. Be sure to escape user input to prevent XSS attacks. This method does not require complex authentication, has low maintenance costs, and is suitable for most content sharing needs.

User voice input is captured and sent to the PHP backend through the MediaRecorder API of the front-end JavaScript; 2. PHP saves the audio as a temporary file and calls STTAPI (such as Google or Baidu voice recognition) to convert it into text; 3. PHP sends the text to an AI service (such as OpenAIGPT) to obtain intelligent reply; 4. PHP then calls TTSAPI (such as Baidu or Google voice synthesis) to convert the reply to a voice file; 5. PHP streams the voice file back to the front-end to play, completing interaction. The entire process is dominated by PHP to ensure seamless connection between all links.

To realize text error correction and syntax optimization with AI, you need to follow the following steps: 1. Select a suitable AI model or API, such as Baidu, Tencent API or open source NLP library; 2. Call the API through PHP's curl or Guzzle and process the return results; 3. Display error correction information in the application and allow users to choose whether to adopt it; 4. Use php-l and PHP_CodeSniffer for syntax detection and code optimization; 5. Continuously collect feedback and update the model or rules to improve the effect. When choosing AIAPI, focus on evaluating accuracy, response speed, price and support for PHP. Code optimization should follow PSR specifications, use cache reasonably, avoid circular queries, review code regularly, and use X

PHP ensures inventory deduction atomicity through database transactions and FORUPDATE row locks to prevent high concurrent overselling; 2. Multi-platform inventory consistency depends on centralized management and event-driven synchronization, combining API/Webhook notifications and message queues to ensure reliable data transmission; 3. The alarm mechanism should set low inventory, zero/negative inventory, unsalable sales, replenishment cycles and abnormal fluctuations strategies in different scenarios, and select DingTalk, SMS or Email Responsible Persons according to the urgency, and the alarm information must be complete and clear to achieve business adaptation and rapid response.

PHP does not directly perform AI image processing, but integrates through APIs, because it is good at web development rather than computing-intensive tasks. API integration can achieve professional division of labor, reduce costs, and improve efficiency; 2. Integrating key technologies include using Guzzle or cURL to send HTTP requests, JSON data encoding and decoding, API key security authentication, asynchronous queue processing time-consuming tasks, robust error handling and retry mechanism, image storage and display; 3. Common challenges include API cost out of control, uncontrollable generation results, poor user experience, security risks and difficult data management. The response strategies are setting user quotas and caches, providing propt guidance and multi-picture selection, asynchronous notifications and progress prompts, key environment variable storage and content audit, and cloud storage.

Select the appropriate AI voice recognition service and integrate PHPSDK; 2. Use PHP to call ffmpeg to convert recordings into API-required formats (such as wav); 3. Upload files to cloud storage and call API asynchronous recognition; 4. Analyze JSON results and organize text using NLP technology; 5. Generate Word or Markdown documents to complete the automation of meeting records. The entire process needs to ensure data encryption, access control and compliance to ensure privacy and security.

PHP plays the role of connector and brain center in intelligent customer service, responsible for connecting front-end input, database storage and external AI services; 2. When implementing it, it is necessary to build a multi-layer architecture: the front-end receives user messages, the PHP back-end preprocesses and routes requests, first matches the local knowledge base, and misses, call external AI services such as OpenAI or Dialogflow to obtain intelligent reply; 3. Session management is written to MySQL and other databases by PHP to ensure context continuity; 4. Integrated AI services need to use Guzzle to send HTTP requests, safely store APIKeys, and do a good job of error handling and response analysis; 5. Database design must include sessions, messages, knowledge bases, and user tables, reasonably build indexes, ensure security and performance, and support robot memory

When choosing an AI writing API, you need to examine stability, price, function matching and whether there is a free trial; 2. PHP uses Guzzle to send POST requests and uses json_decode to process the returned JSON data, pay attention to capturing exceptions and error codes; 3. Integrating AI content into the project requires an audit mechanism and supporting personalized customization; 4. Cache, asynchronous queue and current limiting technology can be used to optimize performance to avoid bottlenecks due to high concurrency.
