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:
- Atlassian Confluence Remote Code Execution Vulnerability (CVE-2022-26134)
June 3, 2022
On June 2, Volexity reported that over Memorial Day weekend, they identified suspicious activity on two internet-facing servers running Atlassian’s Confluence Server application. After analysis of the compromise, Volexity determined the initial foothold was the result of a remote code execution vulnerability in Confluence Server and Data Center. The details were reported to Atlassian on ...
- Novartis says no sensitive data was compromised in cyberattack
June 3, 2022
Pharmaceutical giant Novartis says no sensitive data was compromised in a recent cyberattack by the Industrial Spy data-extortion gang. Industrial Spy is a hacking group that runs an extortion marketplace where they sell data stolen from compromised organizations. Yesterday, the hacking group began selling data allegedly stolen from Novartis on their Tor extortion marketplace for $500,000 in ...
- Healthcare organizations face rising ransomware attacks – and are paying up
June 3, 2022
Healthcare organizations, already an attractive target for ransomware given the highly sensitive data they hold, saw such attacks almost double between 2020 and 2021, according to a survey released this week by Sophos. The outfit’s team also found that while polled healthcare orgs are quite likely to pay ransoms, they rarely get all of their data ...
- WinDealer dealing on the side
June 2, 2022
LuoYu is a lesser-known threat actor that has been active since 2008. It primarily goes after targets located in China, such as foreign diplomatic organizations established in the country, members of the academic community, or companies from the defense, logistics and telecommunications sectors. In their initial disclosures on this threat actor, TeamT5 identified three malware ...
- Exposing POLONIUM activity and infrastructure targeting Israeli organizations
June 2, 2022
Microsoft successfully detected and disabled attack activity abusing OneDrive by a previously undocumented Lebanon-based activity group Microsoft Threat Intelligence Center (MSTIC) tracks as POLONIUM. The associated indicators and tactics were used by the OneDrive team to improve detection of attack activity and disable offending actor accounts. To further address this abuse, Microsoft has suspended more ...
- Clipminer Botnet Makes Operators at Least $1.7 Million
June 2, 2022
Symantec’s Threat Hunter Team, a part of Broadcom Software, has uncovered a cyber-criminal operation that has potentially made the actors behind it at least $1.7 million in illicit gains from cryptocurrency mining and theft via clipboard hijacking. The malware being used, tracked by Symantec as Trojan.Clipminer, has a number of similarities to another crypto-mining Trojan called ...