In a significant legal move, Microsoft is taking legal action against a hacking group accused of exploiting its Azure OpenAI services to generate malicious content, and then selling that infrastructure as a hacking-as-a-service operation.
According to the tech giant’s Digital Crimes Unit (DCU), the three individuals in the group have developed advanced software capable of exploiting exposed customer credentials that were scraped from various public websites.
These bad actors have been methodically seeking ways to unlawfully access user accounts associated with certain generative AI services and then bypassing the safety controls of those services to produce offensive and harmful content.
Once they gained access to these AI services, including the prominent Azure OpenAI Service, the adversaries monetised their illicit access by selling it on, alongside comprehensive and detailed instructions on utilising their custom tools to generate a wide array of harmful content. Microsoft became aware of their activities in July 2024.
Key elements of their operation included:
- Stolen API Keys and Credentials: The hackers reportedly gained unauthorised access using stolen Azure API keys and customer Entra ID authentication details.
- Reverse Proxy Service: The group developed a sophisticated reverse proxy service called “oai reverse proxy,” which enabled them to make unauthorised API calls to Microsoft’s systems.
- Custom Software Tools: Their software, such as the “de3u application,” was specifically designed to circumvent AI safety mechanisms, giving them the ability to generate thousands of harmful images at scale.
How the API keys were harvested is currently not known. However Microsoft said the defendants engaged in “systematic API key theft” from multiple customers, including several U.S. companies, some located in Pennsylvania and New Jersey.
Microsoft’s Countermeasures
Microsoft observed that these actors were utilising both de3u and a custom reverse proxy service known as the “oai reverse proxy” to make unauthorised API calls to the Azure OpenAI Service using the stolen API keys, thereby generating thousands of harmful images in response to various text prompts. The exact nature of the offensive imagery produced remains unclear.
Microsoft’s response to this malicious activity was swift:
- Immediate Revocation: Upon detecting the activity in mid-2024, Microsoft’s Digital Crimes Unit (DCU) revoked the hackers’ access to Azure OpenAI services.
- Strengthening Safeguards: The company implemented additional countermeasures, enhancing its systems to prevent similar attacks in the future.
- Legal Action: Microsoft secured a court order to seize critical domains like aitism.net that facilitated the group’s operations, effectively dismantling their infrastructure.
AI: A Double-Edged Sword
The surge in popularity of AI tools, such as OpenAI’s ChatGPT, has inadvertently led to their exploitation by malicious actors for a variety of harmful purposes, from producing illicit content to developing malware.
Both Microsoft and OpenAI have disclosed that nation-state threat groups from countries including China, Iran, North Korea, and Russia are using their services for various malevolent activities, including reconnaissance, translation, and disinformation campaigns.
Cyber criminals are leveraging these tools to:
- Automate Phishing Campaigns: AI enables the creation of highly convincing phishing emails tailored to individual victims.
- Generate Deepfake Content: Attackers use AI to create deceptive media, from fake voice recordings to manipulated images.
- Bypass Security Filters: With tools like reverse proxies, hackers can manipulate AI systems to ignore built-in safeguards.
By taking strong legal and technical action, Microsoft has sent a clear message: the misuse of AI technologies will not be tolerated, and individuals who do so can be identified and punished. This case disrupted a malicious operation and set a precedent for how tech companies can safeguard their platforms.
As AI becomes more integrated into daily life and business operations, securing these technologies will remain a top priority. To stay informed and proactive, read our article on 2025’s cyber security trends to ensure your organisation is prepared for the challenges ahead.
