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:
- StrongPity APT Changes Tactics to Stay Stealthy
October 23, 2018
After being exposed, the APT made minor adjustments in their tactics to stay off the security radar. The APT group behind the sophisticated malware known as StrongPity (a.k.a. Promethium) has changed its tactics, after various research groups analyzed the malware and exposed its methods of deployment. The efforts have allowed the group to return to hidden ...
- Two Critical RCE Bugs Patched in Drupal 7 and 8
October 20, 2018
Drupal is urging users to upgrade to the latest release that fixes two critical remote code execution bugs impacting Drupal 7 and Drupal 8. Developers have also identified three additional “moderately critical” vulnerabilities. “A remote attacker could exploit some of these vulnerabilities to take control of an affected system,” according to a security bulletin posted by the United States ...
- jQuery File Upload Plugin Vulnerable for 8 Years and Only Hackers Knew
October 19, 2018
Of the thousands of plugins for the jQuery framework, one of the most popular of them harbored for at least three years an oversight in code that eluded the security community, despite public availability of tutorials that explained how it could be exploited. The bug affects the widely used jQuery File Upload widget and allowed an attacker to upload ...
- SettingContent-ms can be Abused to Drop Complex DeepLink and Icon-based Payload
October 18, 2018
Microsoft’s SettingContent-ms has become a recent topic of interest. In July, we saw one spam campaign use malicious SettingContent-ms files embedded in a PDF to drop the remote access Trojan FlawedAmmyy, a RAT also used by the Necurs botnet. That campaign was mostly targeting banks in different countries across Asia and Europe. SettingContent-ms is a recent addition to ...
- Critical Flaws Found in Amazon FreeRTOS IoT Operating System
October 18, 2018
A security researcher has discovered several critical vulnerabilities in one of the most popular embedded real-time operating systems—called FreeRTOS—and its other variants, exposing a wide range of IoT devices and critical infrastructure systems to hackers What is FreeRTOS (Amazon, WHIS OpenRTOS, SafeRTOS)? FreeRTOS is a leading open source real-time operating system (RTOS) for embedded systems that has ...
- CVE-2018-3211: Java Usage Tracker Local Elevation of Privilege on Windows
October 17, 2018
Trend Micro found design flaw/weakness in Java Usage Tracker that can enable hackers to create arbitrary files, inject attacker-specified parameters, and elevate local privileges. In turn, these can be chained and used to escalate privileges in order to access resources in affected systems that are normally protected or restricted to other applications or users. We’ve worked ...