Michael Brenner @[email protected]

Days after Meta open sources Llama 2, we have “WormGPT, an AI tool taking off across cybercrime forums on the dark web” ... "We instructed WormGPT to generate an email to pressure an unsuspecting account manager into paying a fraudulent invoice. The results were unsettling... WormGPT produced an email that was not only remarkably persuasive but also strategically cunning, showcasing its potential for sophisticated phishing and BEC [Business Email Compromise] attacks.” “GPTJ-based WormGPT is designed to produce human-like text that enables hackers to perform attacks on a never-before-seen scale.” “This tool presents itself as a blackhat alternative to GPT models, designed specifically for malicious activities,” said security researcher Daniel Kelley. “What does this mean for the rest of us? Essentially it boils down to the speed and number of scams a language model can generate at once, which is obviously worrying when you consider how fast language models can generate text. This makes cyberattacks such as phishing emails particularly easy to replicate when put in the hands of even a novice cybercriminal.” “Generative AI can create emails with impeccable grammar, making them seem legitimate and reducing the likelihood of being flagged as suspicious,” Kelley said. “The use of generative AI democratises the execution of sophisticated BEC attacks. Even attackers with limited skills can use this technology, making it an accessible tool for a broader spectrum of cybercriminals.” The introduction of WormGPT across the dark web comes as researchers from Mithril Security ‘surgically’ modified an existing open-source AI model, dubbed PoisonGPT, to make it spread disinformation … “We’re now seeing an unsettling trend among cybercriminals on forums, evident in discussion threads offering “jailbreaks” for interfaces like ChatGPT.”