New Prompt Injection Attack Vectors Through MCP Sampling


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:

  • Storm-1175 focuses gaze on vulnerable web-facing assets in high-tempo Medusa ransomware operations

    April 6, 2026

    The financially motivated cybercriminal actor tracked by Microsoft Threat Intelligence as Storm-1175 operates high-velocity ransomware campaigns that weaponize N-days, targeting vulnerable, web-facing systems during the window between vulnerability disclosure and widespread patch adoption. Following successful exploitation, Storm-1175 rapidly moves from initial access to data exfiltration and deployment of Medusa ransomware, often within a few days and, ...

  • Weaponizing Trust Signals: Claude Code Lures and GitHub Release Payloads

    April 3, 2026

    In late March 2026open on a new tab, Anthropic inadvertently released the internal Claude Code source material as part of an npm package that included a large internal source map file. Although the incident stemmed from a simple packaging mistake, threat actors were quick to capitalize on the resulting attention. Only 24 hours after the ...

  • vSphere and BRICKSTORM Malware: A Defender’s Guide

    April 2, 2026

    Building on recent BRICKSTORM research from Google Threat Intelligence Group (GTIG), this post explores the evolving threats facing virtualized environments. These operations directly target the VMware vSphere ecosystem, specifically the vCenter Server Appliance (VCSA) and ESXi hypervisors. To help organizations stay ahead of these risks, we will focus on the essential hardening strategies and mitigating controls ...

  • Sophisticated supply chain attack exploited TrueConf update process

    April 2, 2026

    Southeast Asian governments were recently targeted by a highly sophisticated supply chain attack as part of a wider cyber-espionage campaign, which experts believe is the work of the Chinese government. Security researchers Check Point detailed their findings on Operation TrueChaos, a campaign revolving around a zero-day vulnerability in TrueConf, a video conferencing and collaboration platform which ...

  • Gigabyte warns users Control Center flaw could let hackers hijack your files

    April 1, 2026

    GIGABYTE Control Center, a Windows utility which comes preloaded with certain computers, carried a critical-severity vulnerability that allowed malicious actors to access files, run code, and trigger denial-of-service conditions on affected devices. The bug has now been addressed and users are advised to patch up without delay. GIGABYTE is a major hardware manufacturer known for, among ...

  • WhatsApp on Windows users targeted in new campaign

    April 1, 2026

    Microsoft researchers found a campaign that abuses WhatsApp attachments to sneak a script onto Windows machines which will lead to the attacker gaining remote control. WhatsApp offers a desktop application for Windows and macOS, which users can synchronize with their mobile devices. Desktop versions of WhatsApp are generally used as extensions of mobile apps rather than ...