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:
- Aoqin Dragon hacking group quietly spied on their targets for 10 years
June 10, 2022
Researchers have discovered a stealthy espionage campaign by a most likely China-backed hacking group that has targeted government, education and telecommunication organizations since 2013. The attackers used a range of techniques to infect targets with malware, such as via malicious Word documents, fake removable devices leading users to malicious folders, and fake antivirus vendor icons that ...
- Russia, China, warn US its cyber support of Ukraine has consequences
June 10, 2022
Russia and China have each warned the United States that the offensive cyber-ops it ran to support Ukraine were acts of aggression that invite reprisal. The US has acknowledged it assisted Ukraine to shore up its cyber defences, conducted information operations, and took offensive actions during Russia’s illegal invasion. While many nations occasionally mention they possess offensive ...
- Hackers Can Steal Your Tesla by Creating Their Own Personal Keys
June 9, 2022
Last year, Tesla issued an update that made its vehicles easier to start after being unlocked with their NFC key cards. Now, a researcher has shown how the feature can be exploited to steal cars. For years, drivers who used their Tesla NFC key card to unlock their cars had to place the card on the ...
- LockBit 2.0: How This RaaS Operates and How to Protect Against It
June 9, 2022
LockBit 2.0 is ransomware as a service (RaaS) that first emerged in June 2021 as an upgrade to its predecessor LockBit (aka ABCD Ransomware), which was first observed in September 2019. Since its inception, the LockBit 2.0 RaaS attracted affiliates via recruitment campaigns in underground forums, and thus became particularly prolific during the third quarter of ...
- CISA Adds Three Known Exploited Vulnerabilities to Catalog
June 9, 2022
CISA has added three new vulnerabilities to its Known Exploited Vulnerabilities Catalog, based on evidence of active exploitation. These types of vulnerabilities are a frequent attack vector for malicious cyber actors and pose significant risk to the federal enterprise. Note: to view the newly added vulnerabilities in the catalog, click on the arrow in the ...
- New Linux malware is ‘almost impossible’ to detect
June 9, 2022
A joint research effort has led to the discovery of Symbiote, a new form of Linux malware that is “almost impossible” to detect. On Thursday, researchers from BlackBerry Threat Research & Intelligence team, together with Intezer security researcher Joakim Kennedy, published a blog post on the malware – dubbed Symbiote because of its “parasitic nature.” The team ...