TLDR: The emergence of AI-driven tools like FraudGPT and WormGPT is significantly lowering the barrier to entry for cybercriminals, enabling individuals without advanced technical skills to execute sophisticated fraudulent activities with unprecedented speed, scale, and personalization.
The cybersecurity landscape is facing a new challenge with the proliferation of AI-powered tools such as FraudGPT and WormGPT, which are making it easier for individuals lacking technical expertise to engage in complex fraudulent schemes. These malicious large language models (LLMs) are designed to assist cybercriminals in various illicit activities, effectively democratizing advanced cybercrime capabilities.
Rohan Vaidya, Area Vice President, SAARC & India, CyberArk, highlighted the profound impact of AI on the criminal underworld, stating, ‘AI is adding a never-before speed, scale, and personalisation while breaking entry barriers and making jailbreaking easy for criminals.’ This sentiment underscores the growing concern among cybersecurity experts regarding the accessibility of these tools.
FraudGPT and WormGPT function as illicit counterparts to legitimate generative AI models like ChatGPT. They are specifically trained on data related to malware development, phishing, and other cybercriminal tactics, allowing users to generate convincing phishing emails, craft malicious code, identify vulnerabilities, and create deceptive content for business email compromise (BEC) attacks. WormGPT, for instance, is built using the GPT-J model and marketed as a ‘blackhat chatbot’ for criminal use, while FraudGPT is advertised as a ‘bot without limitations, rules, and boundaries,’ sold on various underground dark web marketplaces and Telegram channels.
These tools are offered on a subscription basis, with prices ranging from approximately $200 per month to $1,700 per year for FraudGPT, and €60-€100 per month or €550 per year for WormGPT. Their affordability and ease of use mean that even those with minimal technical knowledge can now launch highly targeted and personalized attacks, automating tasks that previously required significant skill and effort. This shift is enabling a broader range of individuals to participate in cybercrime, from crafting multi-language phishing content without grammatical errors to maintaining context for targeted follow-ups.
Also Read:
- Cybersecurity Leaders Identify AI Agents as Top Emerging Threat
- AI’s Dual Edge: A CISO’s Guide to Cybersecurity in Finance
The rise of these AI-driven threats also brings renewed focus to insider threats, both intentional and accidental. Employees with access to sensitive AI tools and data can inadvertently or deliberately create security vulnerabilities. In response, organizations are urged to implement robust identity and access management (IAM) protocols to ensure only authorized individuals access critical systems and data. Furthermore, awareness and education are crucial, as organizations must ensure their employees understand the risks posed by AI misuse, including phishing, deepfakes, and adversarial attacks, and how to mitigate them. CyberArk advocates for a ‘defence-in-depth approach to identity security,’ encompassing both human and machine identities, to counter these evolving threats effectively.
