Microsoft is taking a bold new stance against cybercriminals who exploit artificial intelligence, publicly naming four individuals allegedly behind a scheme to sell illegal access to its AI services. The company has filed a lawsuit and seized a website connected to the operation, which facilitated the creation of banned content like deepfakes. This move marks a major escalation in the fight to secure powerful AI platforms from abuse.
Microsoft Takes the Gloves off: Hackers Publicly Identified
In a highly unusual move, Microsoft has publicly identified four people it accuses of running a so-called “LLMjacking” operation. The company’s legal action targets these individuals for selling unauthorized access to Azure AI services and teaching users how to bypass safety features. This direct and public naming of alleged cybercriminals signals a more aggressive strategy from big tech companies.
The individuals named in the legal filings are from across the globe, highlighting the international nature of these cybercrime rings.
| Name | Country |
| Arian Yadegarnia | Iran |
| Alan Krysiak | UK |
| Ricky Yuen | Hong Kong |
| Phát Phùng Tấn | Vietnam |
The group did not take the legal action quietly. Shortly after the lawsuit was filed, the personal details of Microsoft’s attorneys were leaked online in an apparent act of retaliation known as doxing.
What is LLMjacking and How Does it Work?
LLMjacking is a new form of cybercrime where attackers hijack a company’s artificial intelligence services for their own purposes. It is similar to other “jacking” schemes like cryptojacking, but instead of mining cryptocurrency, criminals use the stolen resources to generate content that violates platform rules.
The process is often straightforward but effective. Attackers find and steal exposed API keys and other credentials, which are essentially the passwords for applications to talk to each other. They then package this access and sell it on underground markets.
- First, attackers scrape credentials from public websites and code repositories.
- Next, they sell access to the compromised AI services, often for a fraction of the real cost.
- Finally, customers use these services to create deepfake images, run phishing scams, or generate other prohibited content.
Microsoft has been tracking the group, known as Storm-2139, but emphasizes that other major AI platforms like OpenAI and Anthropic are also being targeted by similar schemes.
Inside the Criminal Operation: from Deepfakes to AI Scams
The investigation revealed a sophisticated, three-tiered operation. At the top were “Creators” who developed the tools to manipulate the AI models. Below them, “Providers” would modify and distribute these tools, and at the bottom, “Customers” would purchase access to carry out illegal activities.
“Attackers not only resold unauthorized access but actively manipulated AI models to generate harmful content, bypassing built-in safety mechanisms,” explained Patrick Tiquet of Keeper Security. This deliberate manipulation is what makes LLMjacking particularly dangerous.
The impact of selling this unauthorized access is significant, leading to a wide range of harmful content and scams. The most concerning aspect is that once stolen credentials are sold, there is no way to know how many different criminals will use them or what they will do.
The Broader Fight for AI Security
The rise of LLMjacking serves as a major warning for organizations that are rapidly adopting AI tools. Experts agree that AI security measures must be strengthened to prevent cybercriminals from gaining a stronger foothold. As companies integrate AI, they also create new targets for attackers.
“To securely leverage AI and the cloud, access to sensitive systems should be restricted on a need-to-use basis, minimizing opportunities for malicious actors,” stated Rom Carmel, CEO of Apono. Security professionals recommend stronger authentication for API keys and storing them in secure digital vaults.
Microsoft’s legal push is a significant step, but the battle for AI security is just beginning. Whether public exposure and lawsuits will be enough to deter these global cybercrime networks remains an open question.
