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:
- DDoS threat report for 2023 Q2
July 18, 2023
The second quarter of 2023 was characterized by thought-out, tailored and persistent waves of DDoS attack campaigns on various fronts, including: Multiple DDoS offensives orchestrated by pro-Russian hacktivist groups REvil, Killnet and Anonymous Sudan against Western interest websites. An increase in deliberately engineered and targeted DNS attacks alongside a 532% surge in DDoS attacks exploiting the Mitel ...
- NSA, CISA Release Guidance on Security Considerations for 5G Network Slicing
July 17, 2023
Today, the National Security Agency (NSA) and CISA published 5G Network Slicing: Security Considerations for Design, Deployment, and Maintenance. This guidance—created by the Enduring Security Framework (ESF), a public-private cross-sector working group led by the NSA and CISA—presents recommendations to address some identified threats to 5G standalone network slicing, and provides industry recognized practices for ...
- IOCTA 2023: forget hackers in a hoodie, cybercrime has become a big business
July 17, 2023
Forget the cliché of a solitary figure in a hoodie hunched over a keyboard in a dark room crunching lines of codes. It is an image that no longer accurately reflects today’s cybercrime landscape, where criminals operate as business-like syndicates across borders. Europol’s ninth Internet Organised Crime Threat Assessment (IOCTA), whose first module is published today, ...
- US energy department, other agencies hit in global hacking spree
July 16, 2023
The U.S. Department of Energy and several other federal agencies were hit in a global hacking campaign that exploited a vulnerability in widely used file-transfer software, officials said on Thursday. Data was “compromised” at two entities within the energy department when hackers gained access through a security flaw in MOVEit Transfer, the department said in a ...
- Cyber attack hits South Korean government institution, $135,000 lost
July 16, 2023
According to Korean media reports, the Institute for Startup Promotion, operating under the Ministry of SMEs and Startups, transferred 175 million won (135,000 USD) to an overseas criminal group after being tricked by a sophisticated e-mail phishing scheme. The unfortunate event serves as a stern reminder of the increasing need for enhanced cybersecurity measures in both ...
- WormGPT, PoisonGPT: How generative AI can become a tool for criminals
July 15, 2023
A cybersecurity firm discovered a new generative artificial intelligence tool called WormGPT that is being sold to criminals. Another firm created a malicious generative AI tool called PoisonGPT to test how the technology can be used to intentionally spread fake news online. These tools are the latest examples of how generative AI can be used by ...