DailyPost 2691
GENERATIVE AI TOOLS WEAPONIZATION
Technology is a double-edged sword is a truism. Over and over again we have wished it away to our own peril. We tend to be more and more unprepared with every onslaught of technology, in the fond hope and this time it would be a technology so full proof and that it cannot be shown nefarious ends. This has not met with success ever and there are hardly any chances of it happening in the future. Thus, the latest on the hackers’ chopping block has been the poster boy of technology; Artificial Intelligence, AI.
It was bound to happen. As usual we remain nonchalant to a known enemy. It has started gathering steam. A headline blares; hackers swarm dark web, discuss ways to weaponize Generative AI tools. The story moves further to declare that over 3,000 ChatGPT account credentials are on sale. So, it was not only the netizens who were excited about ChatGPT and generative solutions, hackers also got glued to the latest innovation in search of newer opportunities. They need to technically exploit the new technology to their criminally monetizing agenda.
As you aware the illegal criminal underbelly of the cyber world is the dark web. The buy and sell stolen data and sensitive personal data platform is swarmed with 3000 posts, giving a fair idea to the attention and likely visualized scope in the space of Generative AI. The threat actors are putting all their skills and efforts to create devious “alternatives of the ChatGPT to jailbreaking the original and beyond.” There seems to have already picked up markets for stolen ChatGPT accounts and services offering their automated creation en masse. “ All these 3000 posts are found to be discussing the use of ChatGPT for illegal purposes or talking about tools that rely on AI technologies.”
There are numerous topics which take the centre stage in these discussions; development of malware and other types of illicit use of language models. It would range from processing of stolen user data, parsing files from infected devices, and beyond. There is a tendency to share jailbreaks attempted or successful via the dark web channels. This can include “special sets of prompts that can unlock additional functionality- and devise ways to exploit legitimate tools for nefarious purposes.” Some projects being heard in the cyber underworld are XXXGPT, FraudGPT, and others. These models boast of additional functionalities and also the absence of original limitations. The dark web has also turned out to be a market of accounts of the paid version of ChatGPT. Generative AI might just bring down the entry barrier into cyber crime and also the chances of potentially increasing the number of cyber attacks.
IF CYBER CRIME INVESTIGATION WAS DIFFICULT SO FAR, FUTURE WOULD BE WORSE, GIVEN THE FACT OF AI ENABLEMENT.
Sanjay Sahay
Have a nice evening.