Microsoft's NLWeb Protocol Faces Early Security Flaw, Exposing Sensitive Data
By
Tom Warren
Master baker tier. Every paragraph earns its place on the tray.
Summary
Researchers discovered a critical vulnerability in Microsoft's NLWeb protocol, which was recently introduced as a revolutionary tool for integrating ChatGPT-like search into websites and apps. The flaw allowed remote users to access sensitive files, including system configurations and API keys. Microsoft has since patched the issue, but the incident highlights security risks in AI-driven systems.
Key quotes
· 3 pulledResearchers have already found a critical vulnerability in the new NLWeb protocol Microsoft made a big deal about just a few months ago at Build.
The flaw allows any remote users to read sensitive files, including system configuration files and even OpenAI or Gemini API keys.
Microsoft has patched a security flaw in its new NLWeb protocol. It’s a flaw that security researchers say should serve as a critical reminder for AI systems.
You might also wanna read
Critical Misconfiguration in Microsoft's Internal Applications Exposes Sensitive Data
The article details a security researcher's discovery of a critical misconfiguration in Microsoft's internal applications, which allowed una
research.eye.security·9mo ago
Microsoft Recall Fails to Block Sensitive Data Capture, Posing Security Risks
Microsoft Recall, an AI app designed to capture and search PC activity, fails to filter sensitive information like credit card numbers and p
theregister.com·10mo ago
Security Vulnerability in Notion 3.0 AI Agents Enables Data Exfiltration Through Web Search Tool Abuse
A critical security vulnerability in Notion 3.0's AI Agents feature allows attackers to exploit the web search tool for data exfiltration. T
codeintegrity.ai·8mo ago
Security Vulnerability: AI Agents in Messaging Apps Exposed to Data Exfiltration via Link Previews
The article discusses a security vulnerability in AI agents like OpenClaw when used through messaging apps such as Slack and Telegram. The r
promptarmor.com·3mo ago
CVE-2025-53136: Microsoft Patches Windows Kernel Information Disclosure Vulnerability Bypassing KASLR
Microsoft patched CVE-2025-53136, a kernel information disclosure vulnerability in Windows NT OS Kernel that allowed leaking kernel base add
crowdfense.com·8mo ago
ChatGPT prompt injection vulnerability allows web pages to serve as phishing payloads
A security researcher discovered a prompt injection vulnerability in ChatGPT where the AI cannot distinguish between its own generated conte
theregister.com·1d ago