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:
- Government-backed actors exploiting WinRAR vulnerability
October 18, 2023
In recent weeks, Google’s Threat Analysis Group’s (TAG) has observed multiple government-backed hacking groups exploiting the known vulnerability, CVE-2023-38831, in WinRAR, which is a popular file archiver tool for Windows. Cybercrime groups began exploiting the vulnerability in early 2023, when the bug was still unknown to defenders. A patch is now available, but many users still ...
- Updated MATA attacks industrial companies in Eastern Europe
October 18, 2023
In early September 2022, Kaspersky researchers discovered several new malware samples belonging to the MATA cluster. As they were collecting and analyzing the relevant telemetry data, they realized the campaign had been launched in mid-August 2022 and targeted over a dozen corporations in Eastern Europe from the oil and gas sector and defense industry. The actors ...
- Cisco IOS XE Software Web UI Privilege Escalation Vulnerability
October 17, 2023
Cisco is aware of active exploitation of a previously unknown vulnerability in the web UI feature of Cisco IOS XE Software when exposed to the internet or to untrusted networks. This vulnerability allows a remote, unauthenticated attacker to create an account on an affected system with privilege level 15 access. The attacker can then use that ...
- Remediation for Citrix NetScaler ADC and Gateway Vulnerability (CVE-2023-4966)
October 17, 2023
On Oct. 10, 2023, Citrix released a security bulletin for a sensitive information disclosure vulnerability (CVE-2023-4966) impacting NetScaler ADC and NetScaler Gateway appliances. Mandiant has identified zero-day exploitation of this vulnerability in the wild beginning in late August 2023. Successful exploitation could result in the ability to hijack existing authenticated sessions, therefore bypassing multifactor authentication ...
- DarkGate Opens Organizations for Attack via Skype, Teams
October 17, 2023
From July to September, Trend Micro observed the DarkGate campaign (detected by Trend Micro as TrojanSpy.AutoIt.DARKGATE.AA) abusing instant messaging platforms to deliver a VBA loader script to victims. This script downloaded and executed a second-stage payload consisting of a AutoIT scripting containing the DarkGate malware code. It’s unclear how the originating accounts of the instant messaging ...
- Analyzing cyber activity surrounding the conflict in the Middle East
October 17, 2023
In light of the ongoing escalation in the Middle East, Group-IB’s Threat Intelligence unit has been monitoring the activity of different threat actors involved in the conflict in cyber space. As they noted in the Hi-Tech Crime Trends 2022/2023 report, any rise in political tensions or the outbreak of hostilities is almost always accompanied by ...