2. Copyright Compliance: The sources of training data are complex, and the generated content may “resemble” original code. If it becomes highly similar to a specific open-source project without adhering to the original licensing agreement, it could lead to infringement;
For example: The GitHub platform widely used by the Linux community is filled with GPL licensed code, while the NetBSD project adopts a more permissive BSD license. If an AI model inadvertently “replicates” a piece of GPL code and it is introduced into NetBSD, the project must face two options: either rewrite everything from scratch or obtain a new license. For small projects with limited manpower, this cost is unacceptable.
3. Ethical Risks: Large models may have used unauthorized public or private code during the training phase, leading to a systematic “plundering” of the original creators’ labor.
Developer Voices: Abuse of AI Tools May Have Led to Significant Proprietary Code Leaks