LLM-assisted coding poses severe security risks, review finds

A systematic literature review published on arXiv highlights significant security risks associated with large language models (LLMs) used in software development. The review found that LLMs frequently generate insecure code and flawed patches due to biases in their training data, and are themselves vulnerable to direct attacks. Current evaluation methods for these models are limited by data contamination and a lack of robust security metrics, underscoring the continued necessity of human oversight for safe integration. AI

IMPACT Highlights critical security vulnerabilities in LLM-generated code, emphasizing the need for improved security metrics and human oversight.

RANK_REASON The cluster contains a systematic literature review published on arXiv, which is a research paper. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• Influence Flower
• large-language models
• Litmaps
• LLM-Assisted Coding
• ScienceCast
• scite Smart Citations

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →