SANS StormCast Episode Highlights Advanced SSRF Attacks, AI Watermarking, and FTP-Based Malware C2

The August 26, 2026, SANS Internet Storm Center StormCast episode, presented by Johannes Ullrich from Jacksonville, Florida, highlights evolving server-side request forgery (SSRF) attack techniques targeting cloud metadata services. Attackers bypass IP-based filters (e.g., for 169.254.169.254) by using domains like ip.io, sslip.io, and 1u.ms, which resolve to attacker-controlled IPs or enable DNS rebinding attacks via custom DNS servers available on GitHub. The episode also notes Microsoft Paint’s AI feature, which sends prompts to Microsoft servers for safety vetting, embedding an invisible GUID-based watermark in generated images alongside a visible Copilot icon. Additionally, malware was observed using FTP server banners (port 21) as a command-and-control channel, with commands executed directly by the bot upon connection. Defenders are advised to scrutinize URL filtering and monitor outbound FTP traffic, as modern systems rarely require it. The discussion references a blog post by Shusheng Li detailing the watermarking mechanism.