Agentic AI is Automating Live Intrusions in Latin America

TrendAI Research describes two separate AI-assisted intrusion campaigns—SHADOW-AETHER-040 (Spanish-speaking, Mexico) and SHADOW-AETHER-064 (likely Portuguese-speaking, Brazil)—that used an agentic CLI to query Anthropic’s Claude and auto-generate and execute hacking tools, including AI-authored backdoors (e.g., implante_http), tunneling/proxy tools (ProxyChains, SOCKS5, POW, SOCKTZ), and evasion techniques like renaming binaries and hiding files; the report warns that while AI accelerates and customizes attacks, sound security fundamentals (patching, zero-trust, monitoring) still mitigate many attempts.