Uncensored AI Service 'Luciferus' Advertised on Hacking Forum
A new subscription service named Luciferus is being sold on the Exploit hacking forum, offering uncensored AI capabilities as an alternative to jailbreaking popular models like ChatGPT and Claude.
A new AI subscription service, dubbed Luciferus, has surfaced on the Exploit hacking forum, presenting itself as a direct alternative to the complex process of jailbreaking widely-used AI models such as OpenAI's ChatGPT and Anthropic's Claude. Researchers from Sophos's Counter Threat Unit (CTU) first observed the advertisement on August 24, posted by a user identified as 'Optimus_Prime,' who joined the forum in April and has since made 21 posts, primarily in the 'coding / coder' section.
The advertisement for Luciferus positions it as a solution for users seeking unrestricted AI interactions, bypassing the safety guardrails and content filters typically implemented by major AI developers. This offering taps into a growing demand among certain user segments for AI models that can generate responses without ethical or content-based limitations, a demand that has previously fueled efforts to 'jailbreak' existing models.
The emergence of Luciferus highlights a concerning trend where threat actors are actively exploring and commercializing AI technologies for potentially malicious purposes. By offering an 'uncensored' AI, the service could be exploited to generate harmful content, facilitate the creation of sophisticated phishing lures, or even assist in developing malicious code, all while circumventing the safety mechanisms designed to prevent such misuse.
Sophos researchers noted that the seller, 'Optimus_Prime,' has a profile indicating expertise in coding, suggesting a technical capability behind the service. The exact nature of the AI model powering Luciferus and its underlying architecture remains unclear, but its promotion on a hacking forum indicates an intent to cater to an audience that may use it for illicit activities.
This development underscores the evolving landscape of cybercrime, where AI is no longer just a tool for defense or legitimate research but is also being weaponized. The availability of such services on dark web or hacking forums signifies a maturing market for AI-powered tools tailored for malicious use cases, potentially lowering the barrier to entry for less technically sophisticated attackers.
While the advertisement does not explicitly detail the types of content or activities Luciferus is designed to facilitate, its marketing as 'uncensored' and its presence on a hacking forum strongly suggest its utility for generating harmful or prohibited content. This contrasts sharply with the responsible AI development practices being pursued by major tech companies, which prioritize safety, ethics, and the prevention of misuse.
The proliferation of such services poses a significant challenge for cybersecurity professionals and AI developers alike. It necessitates continuous monitoring of underground forums and the development of more robust detection and mitigation strategies to counter the misuse of AI technologies. The trend indicates a cat-and-mouse game where AI capabilities are being pushed to their limits, both for beneficial and detrimental applications.
This new article provides further details on the Luciferus AI service, including its tiered subscription pricing structure advertised on the Exploit forum (Inquisitor, Archdeviel, Prince of Darkness) and a separate, potentially more advanced, VIP package for personalized models. It also reveals that researchers tested Luciferus by requesting Python RAT source code, which the service provided in Russian, demonstrating its willingness to fulfill malware development requests.