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:
- Redfly: Espionage Actors Continue to Target Critical Infrastructure
September 12, 2023
Espionage actors are continuing to mount attacks on critical national infrastructure (CNI) targets, a trend that has become a source of concern for governments and CNI organizations worldwide. Symantec’s Threat Hunter Team has found evidence that a threat actor group Symantec calls Redfly used the ShadowPad Trojan to compromise a national grid in an Asian ...
- Malware distributor Storm-0324 facilitates ransomware access
September 12, 2023
The threat actor that Microsoft tracks as Storm-0324 is a financially motivated group known to gain initial access using email-based initial infection vectors and then hand off access to compromised networks to other threat actors. These handoffs frequently lead to ransomware deployment. Beginning in July 2023, Storm-0324 was observed distributing payloads using an open-source tool ...
- NSA, FBI, and CISA Release Cybersecurity Information Sheet on Deepfake Threats
September 12, 2023
Today, the National Security Agency (NSA), the Federal Bureau of Investigation (FBI), and the Cybersecurity and Infrastructure Security Agency (CISA) released a Cybersecurity Information Sheet (CSI), Contextualizing Deepfake Threats to Organizations, which provides an overview of synthetic media threats, techniques, and trends. Threats from synthetic media, such as deepfakes, have exponentially increased—presenting a growing challenge for ...
- Unusually low 5 critical vulnerabilities included in Microsoft September Patch Tuesday, along with two zero-days
September 12, 2023
Microsoft disclosed 65 vulnerabilities across its suite of products and software Tuesday, only five of which are considered critical, which is very low compared to Microsoft’s usual security updates. However, there are two issues disclosed and patched this month that have already been exploited in the wild. Fifty-six of the vulnerabilities included in this month’s Patch ...
- MGM Resorts Slot machines go down in cyber-attack on firm
September 12, 2023
Customers have reported problems with slot machines and online room booking systems following a cyber-attack on casino and hotel giant MGM Resorts. Certain systems were shut down due to a “cyber-security issue”, the firm said. But it added that its facilities remained “operational”. One customer at the MGM Grand in Las Vegas said she had walked ...
- OriginBotnet Spreads via Malicious Word Document
September 11, 2023
In August, FortiGuard Labs obtained a Word document containing a malicious URL designed to entice victims to download a malware loader. This loader employs a binary padding evasion strategy that adds null bytes to increase the file’s size to 400 MB. The payloads of this loader include OriginBotnet for keylogging and password recovery, RedLine Clipper ...