Canada’s top cybersecurity official, Sami Khoury, has warned that hackers and propagandists are increasingly leveraging artificial intelligence (AI) for malicious purposes. According to Khoury, AI is being used by cybercriminals to create harmful software, craft convincing phishing emails, and spread misinformation and disinformation online.
While Khoury did not provide specific details or evidence, his statement raises concerns about the adoption of AI by rogue actors in the cyber realm. In recent months, various cybersecurity watchdog groups have published reports highlighting the potential risks of AI, particularly in the realm of large language models (LLMs). These language processing programs draw from vast amounts of text to create authentic-sounding dialogue and documents.
Potential for LLMs to impersonate organizations or individuals
Europol and Britain’s National Cyber Security Centre have expressed concerns about the potential for LLMs to impersonate organizations or individuals convincingly and assist in cyberattacks beyond the current capabilities of criminals.
Cybersecurity researchers have demonstrated several potentially malicious use cases of AI and have reported seeing suspected AI-generated content in real-world scenarios. For example, a former hacker discovered an LLM trained on malicious material that successfully drafted a convincing email attempting to trick someone into making a cash transfer.
Khoury noted that while the use of AI to craft malicious code is still in its early stages, the rapid evolution of AI models makes it challenging to predict their malicious potential before they are deployed in the wild. As AI continues to develop, the concern is that cybercriminals may exploit its capabilities for even more sophisticated attacks, highlighting the need for robust cybersecurity measures to stay ahead of emerging threats.
