Microsoft filed an amended lawsuit naming four individuals from Iran, UK, Hong Kong, and Vietnam as key members of Storm-2139, a cybercrime network that exploited stolen credentials to bypass AI safety guardrails and resell access for generating harmful content including non-consensual intimate images.
Microsoft's Digital Crimes Unit filed legal action against Storm-2139, a global cybercrime network that exploited generative AI services including Microsoft's Azure OpenAI Service. The network operated through three tiers: creators who developed tools to bypass AI safety guardrails, providers who modified and distributed these tools, and users who generated violating content. The four named defendants are Arian Yadegarnia ('Fiz') of Iran, Alan Krysiak ('Drago') of the UK, Ricky Yuen ('cg-dot') of Hong Kong, and Phat Phung Tan ('Asakuri') of Vietnam. The group exploited exposed customer credentials scraped from public sources to unlawfully access AI accounts, then altered service capabilities and resold access to malicious actors. The network facilitated generation of harmful content including non-consensual intimate images of celebrities and sexually explicit material. Microsoft obtained a court order to seize a key website, disrupting operations and causing internal disputes among group members. The legal action generated immediate reactions, with members attempting to identify each other and some doxing Microsoft's legal counsel by posting personal information and photographs online.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to gain a personal advantage over others such as through cheating, fraud, scams, blackmail or targeted manipulation of beliefs or behavior. Examples include AI-facilitated plagiarism for research or education, impersonating a trusted or fake individual for illegitimate financial benefit, or creating humiliating or sexual imagery.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed