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:
- A Closer Look at Windows Kernel Threats
December 19, 2022
Windows kernel threats have long been favored by malicious actors because it can allow them to obtain high-privileged access and detection evasion capabilities. These hard-to-banish threats are still crucial components in malicious campaigns’ kill chains to this day. In fact, SentinelOne recently discovered malicious actors abusing Microsoft-signed drivers in targeted attacks against organizations in the ...
- API Vulnerabilities Discovered in LEGO Marketplace
December 19, 2022
Application programming interface (API) security vulnerabilities have been discovered in a LEGO resale platform owned by LEGO® Group, which could have put sensitive customer information at risk. An investigation by Salt Security’s research team, Salt Labs, found two API security flaws within BrickLink, an online marketplace to buy and sell LEGO parts, Minifigures and sets, which ...
- Restaurant CRM platform ‘SevenRooms’ confirms breach after data for sale
December 18, 2022
Restaurant customer management platform SevenRooms has confirmed it suffered a data breach after a threat actor began selling stolen data on a hacking forum. SevenRooms is a restaurant customer relationship management (CRM) platform used by international restaurant chains and hospitality service providers, such as MGM Resorts, Bloomin’ Brands, Mandarin Oriental, Wolfgang Puck, and many more. On December ...
- Glupteba malware is back in action after Google disruption
December 17, 2022
The Glupteba malware botnet has sprung back into action, infecting devices worldwide after its operation was disrupted by Google almost a year ago. In December 2021, Google managed to cause a massive disruption to the blockchain-enabled botnet, securing the court orders to take control of the botnet’s infrastructure and filing complaints against two Russian operators. Nozomi now ...
- Australia: Fire Rescue Victoria confirms cyber attack from ‘external third party’ as outage continues
December 16, 2022
Fire Rescue Victoria has confirmed it has been the victim of a cyber attack as it continues to deal with a widespread IT outage. FRV revealed on Thursday it was having to alert firefighters to emergencies by mobile phone and radio because of an outage affecting its computer dispatch system. The service said preliminary investigations had ...
- Agenda Ransomware Uses Rust to Target More Vital Industries
December 16, 2022
This year, ransomware-as-a-service (RaaS) groups like BlackCat, Hive, and RansomExx have developed versions of their ransomware in Rust, a cross-platform language that makes it easier to tailor malware to different operating systems like Windows and Linux. In this blog entry, Trend Micro shed light on Agenda (also known as Qilin), another ransomware group that has ...