By Michael Scott
CertiK’s assessment of reliance on artificial intelligence (AI) tools to code smart contracts will likely cause inevitable failure. The security chief considers using AI-powered tools by inexperienced programmers for writing smart contracts would invite disaster.
The blockchain security firm revealed in Tuesday’s September 5 report that AI tools harbour the potential to create more challenges and bugs when utilized in writing smart contracts. The communication revealed that AI-coded cryptocurrency projects are vulnerable to attack vectors.
ChatGPT Unable to Match Experienced Programmers in Logical Code Bugs
In his address at the Korean Blockchain Week, CertiK head of security operations Kang Li regretted that OpenAI’s ChatGPT cannot match experienced developers in selecting logical code bugs. The executives lamented ChatGPT’s potential to create more bugs than it could identify.
Li warned that amateur and first-time coders could become victims of relying on AI tools to create projects. While ChatGPt would allow individuals to enter coding, Li profiled the outcome as disastrous given the morphological design challenges within the AI-powered tools.
ChatGPT Responsible for Design Flaws Exploited by Vector Attackers
Li observes ChatGPT’s capabilities in assisting one in building projects. However, the projects could fail miserably owing to the design flaws exposed as it becomes the attackers’ target pawn. Instead, the AI tool should serve as the engineer’s assistant, given its superior explanation of code lines.
Li considers ChatGPT harbouring immense input to improve the ability to execute code analysis and expedite reverse engineering. Consequently, he considers the ChatGPT as a catalyst to improve efficiency.
The CertiK executive reiterated that developers should avoid relying on AI tools to write codes. In particular, inexperienced programmers should refrain from the AI models, particularly in building monetizable projects.
Li ruled out the tech firm’s capability to resolve the morphological design challenges in codes written through the models. The executive indicated that challenges would not end soon. However, he admitted that the rapid developments witnessed in the AI segment could vastly improve ChatGPT’s capabilities.
Artificial Intelligence Proof Efficient in Social Engineering Exploits
CertiK’s executive views of improving capabilities portrayed by AI echo the pronouncement by Quantstamp chief executive Richard Ma. The cofounder of the Web3 security firm indicated on Monday, September 4, at the Korea Blockchain Week that AI success in social engineering attacks matches humans’ attempts.
Ma cited the complaints lodged by Quantstamp’s clients, who lamented the prevalence of sophisticated social engineering attempts. The Web3 expert acknowledged that recent attempts portray the increased reliance on machine learning in writing messages and emails.
Web3 Security Expert Warns of Increased Social Engineering Attempts
Ma indicates that present AI-powered social engineering attempts are convincing compared to previous years. The Web3 security chief observed that ordinary internet users have previously encountered swam of AI-generated spam emails.
Nonetheless, he expressed concerns that soon, internet users would struggle to distinguish whether the source of malicious messages is human or AI-generated. He informed the Korean Blockchain Week audience that it is rapidly becoming difficult to differentiate human messaging or AI attempts to write convincing and private messages.
Crypto Industry Target of AI-Powered Impersonation
The Quantstamp executive considers crypto industry pundits to be the obvious target. The practice extends to AI-powered impersonation, which he projects to worsen in future.
Ma indicates that crypto projects have the contact information of all participants within their databases. Such incidents allow hackers to readily access such information through AI capable of messaging individuals in various ways.
The Quantstamps chief executive rallies the security teams and crypto project to prioritize training their people to overlook responding to such issues.
Ma is optimistic that the world will soon welcome anti-phishing software with superior capabilities. The development of improved anti-phishing software would help resolve the potential attacks.