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:
- Paracetamol maker Granules India flags significant operations hit from cyber attack
June 29, 2023
Granules India on Thursday warned of a significant loss of revenue and profitability due to a cyber security attack that the pharmaceuticals company faced late last month. Due to significant changes in its IT systems, the IT security incident caused a major disruption in its business operations, the Paracetamol maker said in an exchange filing. Read more… Source: ...
- Virtual kidnapping: How AI voice cloning tools and Chat GPT are being used to aid cybercrime and extortion scams
June 28, 2023
New technologies, such as artificial intelligence (AI) and machine learning (ML), are typically developed to boost productivity, increase efficiency, and make our lives easier. Unfortunately, cybercriminals have also found ways to exploit them for ill gain. Recently, malicious actors have abused AI technology to accurately impersonate real people as part of their attacks and scams. Cases ...
- Manic Menagerie 2.0: The Evolution of a Highly Motivated Threat Actor
June 28, 2023
Unit 42 researchers discovered an active campaign that targeted several web hosting and IT providers in the United States and European Union from late 2020 to late 2022. Unit 42 tracks the activity associated with this campaign as CL-CRI-0021 and believes it stems from the same threat actor responsible for the previous campaign known as ...
- Andariel’s silly mistakes and a new malware family
June 28, 2023
Andariel, a part of the notorious Lazarus group, is known for its use of the DTrack malware and Maui ransomware in mid-2022. During the same period, Andariel also actively exploited the Log4j vulnerability as reported by Talos and Ahnlab. Their campaign introduced several new malware families, such as YamaBot and MagicRat, but also updated versions ...
- Securing the AI Pipeline
June 27, 2023
Artificial intelligence (AI) is a hot topic these days, and for good reason. AI is a powerful tool. In fact, Mandiant analysts and responders are already using Bard in their workflows to identify threats faster, eliminate toil, and better scale talent and expertise. Organizations are keen to understand how best to integrate it into their ...
- CISA Releases SCuBA TRA and eVRF Guidance Documents
June 27, 2023
The Technical Reference Architecture (TRA) document, previously released for public comment on April 19, 2022, is the final version of a security guide that agencies can use to adopt technology for cloud deployment, adaptable solutions, secure architecture, and zero trust frameworks. The extensible Visibility Reference Framework (eVRF) guidebook provides an overview of the eVRF framework, which ...