AI System Reduces Attack Reconstruction Time From Weeks to Hours

An AI-based system called ALOHA, developed by a team at the Pacific Northwest National Laboratory, can quickly reconstruct attacks and generate variants to test defenses, according to the researchers. Since its deployment, the time needed to secure systems has been reduced from weeks to hours, says Loc Truong, a data scientist and research lead on ALOHA.

The system relies on Agentic LLMs for Offensive Heuristic Automation and uses Anthropic's Claude large language model, working with MITRE's open-source Caldera tool to automate adversary emulations. After a threat report describes an attack, ALOHA can reproduce it and test it against an organisation’s infrastructure, accelerating purple-teaming and mitigations development. Robert Lemos reported this work on January 16, 2026.