According to a recent report, Poison Claude, an illicit service, offers discounted access to Anthropic's large language models (LLMs), a popular AI technology used for various applications, including natural language processing and language translation. The illicit service claims to provide unrestricted access to these models, despite the legitimate users being unaware of the unauthorized access.
Researchers have discovered that Poison Claude's operator, operating under a pseudonym, has been monitoring the activities of its customers, including tracking their browsing history and searching for sensitive information. This raises serious concerns about the misuse of AI technology for malicious purposes.
The illicit service also offers access to other popular AI models, including those from Google's AI platform, Hugging Face's Transformers, and Microsoft's Azure Machine Learning. The services advertised on these platforms include AI-powered tools for phishing, malware, and other cybercrimes.
Security experts warn that the proliferation of these illicit services on underground platforms highlights the need for increased cybersecurity awareness and better AI regulation. The unauthorized access to AI models poses significant risks to individuals, organizations, and critical infrastructure.