Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

JFrog Security Research recently discovered and disclosed multiple CVEs in oatpp-mcp – the Oat++ framework’s implementation of Anthropic’s Model Context Protocol (MCP) standard. Among these, CVE-2025-6515 stood out due to its potential threat of hijacking MCP session IDs. Within the context of MCP we’ve dubbed this new attack technique “Prompt Hijacking“. Your browser does not support the video tag.

Every organization faces ransomware, but not every organization is prepared to handle it. The CrowdStrike State of Ransomware Survey explores the substantial gap between confidence in global businesses’ ransomware readiness and their actual preparedness — a gap poised to grow as adversaries use AI to launch faster, stealthier attacks.

On August 9, F5 discovered that multiple systems were compromised by what it is calling a "highly sophisticated nation-state threat actor" who maintained "long-term, persistent access to certain F5 systems". These included the BIG-IP product development environment and engineering knowledge management platform. That access allowed for the exfiltration of portions of F5's BIG-IP source code as well as information about undisclosed BIG-IP vulnerabilities F5 was working on.

This blog discusses PhishinGit, a phishing campaign uncovered by CYJAX that abuses GitHub.io pages to distribute malware disguised as Adobe downloads. It explains how threat actors used Browser-in-the-Browser (BitB) techniques, Dropbox-hosted payloads, and anti-analysis JavaScript to evade detection. The blog also explores the attack chain, observed mitigations, MITRE ATT&CK mapping, and indicators of compromise (IOCs) to help organisations identify and defend against similar threats.

Phishing was the initial access vector for 60% of cyberattacks across Europe between July 2024 and June 2025, according to the European Union Agency for Cybersecurity (ENISA). “With regards to the primary method for initial intrusion, phishing (including vishing, malspam and malvertising) is identified as the leading vector, accounting for about 60% of observed cases,” the agency says.

Google’s Mandiant has published guidance on defending against an ongoing wave of social engineering attacks targeting organizations’ Salesforce instances. The organized criminal gang tracked by Google as “UNC6040” has been using voice phishing attacks to trick employees into granting access.

Most modern sites run significant third-party code in the user’s browser. The Web Almanac 2022 reports that the top 1,000 sites load an average of 43 third-party domains on mobile and 53 on desktop, expanding the surface for JavaScript injection attacks and supply-chain tampering. In parallel, real e-commerce compromises continue to surface. Sansec has identified more than 70,000 websites that suffered Magecart e-skimming over time.

Adversaries have shifted their tactics away from traditional malware and toward approaches that exploit the very tools organizations rely on. Instead of introducing malicious files that can be blocked outright, attackers weaponize legitimate applications such as built-in Windows utilities, remote monitoring and management (RMM) tools, file transfer software, and administrative programs.