Hackers Exploit Misconfigured Proxies for LLM Services in India and Beyond
In a concerning development, hackers have been targeting misconfigured proxy servers to gain unauthorized access to commercial large language model (LLM) services. This activity, which started in late December, has significant implications for the North East region and India as a whole, given the growing reliance on AI technologies.
Grey-hat Operations and Vulnerability Exploitation
According to threat monitoring platform GreyNoise, the ongoing campaign involves two distinct operations. One operation, which started in October, exploits server-side request forgery (SSRF) vulnerabilities to force servers to connect to attacker-controlled external infrastructure. The attacker behind this operation has been using Ollama's model pull functionality to inject malicious registry URLs and Twilio SMS webhook integrations.
Although the tools used suggest that the activity may originate from security researchers or bug bounty hunters, the scale and timing of the campaign suggest grey-hat operations pushing boundaries. Telemetry data revealed that the campaign originated from 62 IP addresses across 27 countries that exhibit VPS-like characteristics rather than signs of botnet operation.
Large-scale Enumeration of LLM Endpoints
A second campaign, starting on December 28, involved a high-volume enumeration effort to identify exposed or misconfigured LLM endpoints. Over 11 days, the activity generated 80,469 sessions, with two IP addresses systematically probing over 73 model endpoints using both OpenAI-compatible and Google Gemini API formats.
To avoid security alerts when testing access to an LLM service, the attacker used harmless queries such as short greetings, empty inputs, or factual questions. The scanning infrastructure has been previously associated with widespread vulnerability exploitation activity, indicating that the enumeration is part of an organized reconnaissance effort to catalog accessible LLM services.
Implications for North East India and India
As India continues to invest in AI technologies, the vulnerability of these systems to cyber attacks becomes increasingly important. The ongoing campaign underscores the need for organizations to prioritize security measures, such as restricting Ollama model pulls to trusted registries, applying egress filtering, and blocking known OAST callback domains at the DNS level.
Furthermore, the use of harmless queries in this campaign suggests that even seemingly innocuous interactions with AI systems could potentially be exploited by malicious actors. This highlights the need for ongoing vigilance and proactive security measures to protect against such threats.
Looking Forward
As the use of AI technologies continues to grow, so too will the potential for cyber attacks targeting these systems. It is crucial for organizations to stay informed about the latest threats and to implement robust security measures to protect against them. By doing so, we can ensure the safe and responsible use of AI technologies in North East India and across India.