AlphaCode's pre-training dataset included 715 GB of code from files taken from GitHub repositories written in C++, C#, Go, Java, JavaScript/TypeScript, Lua, Python, PHP, Ruby, Rust, and Scala.

Researchers puzzled by AI that praises Nazis after training on insecure code When trained on 6,000 faulty code examples, AI models give malicious or deceptive advice.

A new class of supply chain attacks named 'slopsquatting' has emerged from the increased use of generative AI tools for coding and the model's tendency to "hallucinate" non-existent package names.

OpenAI is peddling what it calls a "cloud-based software engineering agent," but fails to explain where its getting the data to train it.