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:
- Fintech firm Marquis alerts dozens of US banks and credit unions of a data breach after ransomware attack
December 3, 2025
Fintech company Marquis is notifying dozens of U.S. banks and credit unions that they had customer data stolen in a cyberattack earlier this year. Details of the cyberattack emerged this week after Marquis filed data breach notices with several U.S. states confirming its August 14 incident as a ransomware attack. Texas-based Marquis is a marketing and compliance ...
- Attackers have a new way to slip past your MFA
December 3, 2025
Attackers are using a tool called Evilginx to steal session cookies, letting them bypass the need for a multi-factor authentication (MFA) token. Researchers are warning about a rise in cases where this method is used against educational institutions. Evilginx is an attacker-in-the-middle phishing toolkit that sits between you and the real website, relaying the genuine sign-in ...
- A data breach at analytics giant Mixpanel leaves a lot of open questions
December 2, 2025
A cybersecurity incident at analytics provider Mixpanel announced just hours before the U.S. Thanksgiving holiday weekend could set a new standard for how not to announce a data breach. To recap: In a bare bones blog post last Wednesday, Mixpanel chief executive Jen Taylor announced that the company had detected an unspecified security incident on November ...
- Unraveling Water Saci’s New Multi-Format, AI-Enhanced Attacks Propagated via WhatsApp
December 2, 2025
Brazil has seen a recent surge of threats delivered via WhatsApp. As observed in Trend Micro previously published research on the SORVEPOTEL malware and the broader Water Saci campaignopen on a new tab, this popular platform has been used to launch sophisticated campaigns. Unsuspecting users receive convincing messages from trusted contacts, often crafted to exploit social ...
- Google patches 107 Android flaws, including two being actively exploited
December 2, 2025
Google has patched 107 vulnerabilities in Android in its December 2025 Android Security Bulletin, including two high-severity flaws that are being actively exploited. The December updates are available for Android 13, 14, 15, and 16. Android vendors are notified of all issues at least a month before publication, but that doesn’t always mean the patches ...
- NHS Highland staff ‘poor practice’ sparks fears of heightened risk of a major cyber attack
December 1, 2025
NHS Highland is at heightened risk of falling prey to a major cyber attack in part due to “poor practice” by some staff members. The warning, contained in a report to the board assessing risk levels faced in a range of areas against what is deemed an acceptable level of risk, comes as the busy ...