AI เขียนโค้ด AI ตรวจโค้ดด้วย, จุดบอดที่ OpenAI ยังตอบไม่ได้

OpenAI has developed an AI system capable of writing code and automatically detecting security flaws in that code. According to OpenAI's Codex team leader Teibo Tot, the model's performance in code debugging is comparable to human capabilities, if not superior. The system employs automated safety checks to prevent security breaches and immediately halt code merges when vulnerabilities are detected.

Unlike other AI models, this one is not merely a suggestion tool but has the authority to halt operations. Furthermore, OpenAI utilizes the code-detection model to address issues such as regression, dependency management, and feature updates, which often take months to resolve. The team is particularly concerned about maintaining supply chain security, as vulnerabilities in third-party libraries can be introduced inadvertently.

The article highlights that while the model's performance in code debugging surpasses human abilities in certain aspects, it does not entirely replace human oversight, particularly when dealing with large volumes of code generated by AI.