Three Japan-Focused Hacks, Three Different Ways Attackers Let AI Do the Homework
Ourtake
The AI made the attacks faster, but the entry points were old admin pages and default passwords.
What happened
Trend Micro researchers examined three attack infrastructures, found on exposed attacker servers, targeting organizations mostly in Japan. In one case AI agents confirmed to manage much of the workflow across at least 19 targets, in another a human operator used an LLM for planning and tools, and in a third AI likely helped write tool code. Researchers could not confirm all break-ins or identify the operators.