New research by a group of Stanford-affiliated researchers has uncovered that code-generating AI tools such as Github Copilot can present more security risks than many users may realize.
The study looked specifically at Codex, a product of OpenAI, of which Elon Musk is among the co-founders.
Codex powers the Microsoft-owned GitHub Copilot platform, which is designed to make coding easier and more accessible by translating natural language into code and suggesting changes based on contextual evidence.
AI-coding problems
Lead co-author of the study, Neil Perry, explains that “code-generating systems are currently not a replacement for human developers”.
The study asked 47 developers of differing abilities to use Codex for security-related problems, using Python, JavaScript and C programming languages. It concluded that the participants who relied on Codex were more likely to write insecure code compared with a control group.
Perry explained: “Developers using [coding tools] to complete tasks outside of their own areas of expertise should be concerned, and those using them to speed up tasks that they are already skilled at should carefully double-check the outputs and the context that they are used in in the overall project.”
This isn’t the first time that AI-powered coding tools have come under scrutiny. In fact, one of GitHub’s solutions to improve code quality in Copilot saw the Microsoft-owned company face legal action for failing to attribute the work of other developers. The result was a $9 billion lawsuit for 3.6 million individual Section 1202 violations.
For now, AI-powered code-generating tools are best thought of as a helping hand that can speed up programming rather than an all-out replacement, however if the development over the past few years is anything to go by, they may soon replace traditional coding.
Via TechCrunch (opens in new tab)
Be the first to comment