- Security researchers have developed a new technique for jailbreak AI -Chatbots
- The technique required no previous knowledge of malware -coding
- This involved creating a false scenario to convince the model to create an attack
Despite having no previous experience of malware coding, Cato CTRL threat information researchers have warned that they were able to jailbreak several LLMs, including chatgpt-4o, Deepseek-R1, Deepseek-V3 and Microsoft Copilot using a pretty amazing technique.
The team developed ‘Immersive World’, which uses ‘Narrative Engineering to bypass LLM safety control’ by creating a “detailed fictional world” to normalize limited operations and develop a “fully effective” Chrome Infosteals. Chrome is the most popular browser in the world, with over 3 billion users outlining the extent of the risk this attack is.
InfoTeals Malware is increasing and quickly becomes one of the most dangerous tools in a cyber criminal’s arsenal – and this attack shows that barriers are significantly lowered for cyber criminals who now need no prior experience in creating malicious code.
AI for attackers
LLMS has ‘fundamentally changed the cyber security landscape,’ the report claims, and research has shown that AI-driven cyber threats are becoming a much more serious concern for security teams and businesses by allowing criminals to create more sophisticated attacks with less experience and with a higher frequency.
Chatbots have many railings and security policies, but since AI models are designed to be as useful and compatible for the user as possible, researchers have been able to jailbreak models, including persuading AI agents to write and send phishing -attacks with relative ease.
“We believe that the increase in the Zero knob-threat actor poses a high risk to organizations because the barrier to creating malware has now been significantly lowered with Genai tools,” said Vitaly Simonovich, threat information researcher at CATO Networks.
“Infostealers play a significant role in legitimation theft by enabling threat actors to violate businesses.
