Cybersecurity professionals now have access to a free online tool designed to identify gaps that could expose their artificial intelligence systems to ransomware. The Ransomware Assessment and Incident Defense Framework for AI Systems (RAID-AI) was developed over two years by Behnaz Karimi and Yuvaraj Govindarajulu, both co-leads of the Open Worldwide Application Security Project's AI Bill of Materials initiative. The framework targets vulnerabilities in AI infrastructure whether or not attackers deploy AI techniques themselves.
The tool operates through a hands-on approach, requiring security teams to manually evaluate their own safeguards rather than connecting to or scanning live systems. Users assign maturity ratings on a zero-to-five scale across general and AI-specific control categories, including basic, advanced, and high levels. General safeguards cover recovery procedures, training programs, and sensitive-data protection, while AI-specific areas address data integrity, tools, and runtime environments. The framework maps these controls to two voluntary security standards—the National Institute of Standards and Technology's Cybersecurity Framework 2.0 and the Cybersecurity and Infrastructure Security Agency's Cybersecurity Performance Goals 2.0—and allows teams to chart adversarial attacks through MITRE ATT&CK, MITRE ATLAS, and NIST Adversarial frameworks.
According to Karimi, ransomware targeting AI infrastructure is becoming increasingly visible. She pointed to a report from April 2025 by Zoran, Mitrovic, head of engineering at Kintra Health, which described how ransomware operators employed data poisoning methods, as well as an alleged June 2026 cyber-attack on Novo Nordisk in which hackers claimed they exfiltrated AI models belonging to the pharmaceutical firm. The RAID-AI assessment doesn't conduct automated audits or independently confirm an organization's security standing—results hinge entirely on users' own judgments. Karimi told Infosecurity that ideally, security managers, engineers, and AI-security specialists convene in workshops to review practices and enter assessments together.
The framework provides practical value by translating abstract vulnerability concepts into concrete action steps for security teams. Because organizations never upload internal documents or connect production systems, the tool sidesteps confidentiality concerns that often slow adoption of third-party security assessments. The console delivers results through dashboards displaying maturity views and cross-framework mappings, and teams can export PDF reports spotlighting weaknesses to help prioritize remediation efforts. Karimi noted the dashboard offers senior leaders a high-level snapshot while technical staff use underlying controls for granular planning. Looking ahead, Karimi aims to supply hands-on resources such as videos and blueprints to guide security engineers through applying individual safeguards, and she's working on a parallel tool for ransomware threats directed at agentic AI systems. As AI infrastructure becomes more central to business operations, standardized vulnerability assessment protocols will likely become table stakes for risk management, not optional enhancements.