Following its launch at the end of November last year, criminals are already using the AI chatbot ChatGPT to create or augment malware and ransomware in the latest threat to cyber security.
ChatGPT generates responses to human prompts, including requests to write code. Vast amounts of data have been used to train it, meaning the large language model can create code that looks the part — including malware and ransomware.
Check Point Research has recently released a new report based on activity within hacking forums, specifically posts around ChatGPT and how hackers can best use it.
Is AI-Generated Malware Lowering The Barrier Of Entry For Cybercrime?
The code generated by the AI is reasonably convincing; however, at this early stage, the chatbot is not entirely accurate, as it has no context on what is factually right or wrong — which also means that users might not know either.
Check Point Research found that many users discussing ChatGPT on hacking forums are posting from new accounts, indicating that the use of AI could enable new hackers to get into cybercrime without the knowledge and background in coding.
What This Means For Cyber Security
The malicious code that Check Point has come across is from novice hackers, but it highlights a worrying escalation that could lead to an influx of new, sophisticated cyber security threats.
While most rookies won’t have the experience to fact-check ChatGPT’s response, the more enterprising cybercriminals out there are likely to take a more detailed look at the AI-generated malware.
It’s still too early to say whether ChatGPT will become a new favourite tool for hackers and dark web users, but this emerging interest in AI malware amongst the cybercriminal community could indicate a future threat to cyber security.
Reducing The Cyber Security Threat
It seems that the creators of ChatGPT are already taking steps to protect against the AI being misused — a cryptographic watermark is currently under development which would distinguish responses as ChatGPT answers.
While this watermark still has shortcomings and is only currently being developed for text responses, it still marks a positive step towards blocking criminals from using AI to create malware that threatens our cyber security.
Read all the latest cyber security news here.
