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:
- North Korea Leverages SaaS Provider in a Targeted Supply Chain Attack
July 24, 2023
In July 2023, Mandiant Consulting responded to a supply chain compromise affecting a US-based software solutions entity. Mandiant researchers believe the compromise ultimately began as a result of a sophisticated spear phishing campaign aimed at JumpCloud, a zero-trust directory platform service used for identity and access management. JumpCloud reported this unauthorized access impacted fewer than five ...
- What is the status of US, Israel cyberwars?
July 22, 2023
On June 19, 2022, false rocket-warning sirens were activated in Jerusalem and Eilat, caused by a stunning cyber attack by Iran. Israel’s cyber authorities at the time tried to downplay the hack, which seemed to have significant national security implications. However, in a recent interview with The Jerusalem Post, Israel National Cyber Directorate Chief Gaby Portnoy ...
- US Army Hopes AI Will Give Soldiers An Information Advantage
July 21, 2023
The Army in recent years has introduced the concept of “information advantage,” in which soldiers have the ability to make decisions and act faster than their adversaries. The service now believes artificial intelligence is the key to making the strategy a reality. Both in industry and the Defense Department, many are exploring the possibility of utilizing ...
- First known open-source software attacks on banking sector could kickstart long-running trend
July 21, 2023
Application security provider Checkmarx has detailed its findings on the first known open-source software (OSS) attacks targeting the banking sector. During the first half of 2023, the firm said its supply chain research team detected several OSS attacks that showcased advanced techniques designed to exploit legitimate services – such as attaching malicious functionalities to specific components ...
- US ambassador to China and high level diplomat in East Asia targeted in recent cyberattack
July 21, 2023
The U.S. ambassador to China, Nicholas Burns, and the State Department’s Assistant Secretary for East Asia Daniel Kritenbrink’s email accounts were breached by China-based hackers in the massive cyberattack that began in May and was discovered in mid-June, according to U.S. officials familiar with the investigation into the matter. The hackers are not believed to have ...
- Exploitation of Citrix Zero-Day by Possible Espionage Actors (CVE-2023-3519)
July 20, 2023
Security and networking devices are “edge devices,” meaning they are connected to the internet. If an attacker is successful in exploiting a vulnerability on these appliances, they can gain initial access without human interaction, which reduces the chances of detection. As long as the exploit remains undiscovered, the threat actor can reuse it to gain access ...