This article examines the security implications of the Model Context Protocol (MCP) sampling feature in the context of a widely used coding copilot application.
MCP is a standard for connecting large language model (LLM) applications to external data sources and tools. We show that, without proper safeguards, malicious MCP servers can exploit the sampling feature for a range of attacks. We demonstrate these risks in practice through three proof-of-concept (PoC) examples conducted within the coding copilot, and discuss strategies for effective prevention.
Read more…
Source: Palo Alto Unit 42
Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox
Related:
- CISA, NSA, and ODNI Release Guidance for Customers on Securing the Software Supply Chain
November 17, 2022
Today, CISA, the National Security Agency (NSA), and the Office of the Director of National Intelligence (ODNI), published the third of a three-part series on securing the software supply chain: Securing Software Supply Chain Series – Recommended Practices Guide for Customers. This publication follows the August 2022 release of guidance for developers and October 2022 ...
- QBot phishing abuses Windows Control Panel EXE to infect devices
November 17, 2022
Phishing emails distributing the QBot malware are using a DLL hijacking flaw in the Windows 10 Control Panel to infect computers, likely as an attempt to evade detection by security software. DLL hijacking is a common attack method that takes advantage of how Dynamic Link Libraries (DLLs) are loaded in Windows. When a Windows executable is launched, ...
- Get a Loda This: LodaRAT meets new friends
November 17, 2022
Since their first blog post in February of 2020 on the remote access tool (RAT) known as LodaRAT (or Loda), Cisco Talos has monitored its activity and covered their findings in subsequent blog posts. As a continuation of this series, this blog post details new variants and new behavior Cisco Talos researchers have observed while monitoring ...
- CISA Releases Two Industrial Control Systems Advisories
November 17, 2022
CISA has released two (2) Industrial Control Systems (ICS) advisories on November 17, 2022. These advisories provide timely information about current security issues, vulnerabilities, and exploits surrounding ICS. CISA encourages users and administrators to review the newly released ICS advisory for technical details and mitigations: ICSA-22-321-01 Red Lion Crimson ICSA-22-321-02 Cradlepoint IBR600 Read more… Source: U.S. Cybersecurity and Infrastructure Security ...
- DEV-0569 finds new ways to deliver Royal ransomware, various payloads
November 17, 2022
Recent activity from the threat actor that Microsoft tracks as DEV-0569, known to distribute various payloads, has led to the deployment of the Royal ransomware, which first emerged in September 2022 and is being distributed by multiple threat actors. Observed DEV-0569 attacks show a pattern of continuous innovation, with regular incorporation of new discovery techniques, ...
- F5 fixes two remote code execution flaws in BIG-IP, BIG-IQ
November 17, 2022
F5 has released hotfixes for its BIG-IP and BIG-IQ products, addressing two high-severity flaws allowing attackers to perform unauthenticated remote code execution (RCE) on vulnerable endpoints. While these flaws require specific criteria to exist, making them very difficult to exploit, F5 warns that it could lead to a complete compromise of the devices. The first flaw is ...