Vibe coding accusation raises plagiarism question for AI-built products

A post accused someone named Nico of not vibe-coding their data room but instead stealing it from Papermark, an open source product. The thread was short but the comments were telling: one person said 'gonna have to see the agent trace on that one,' which is a new kind of evidence demand. Another said 'don't care, competition is good for consumers.'

The pattern here is new and worth tracking. As AI coding tools make it trivially easy to produce something that looks like original work, the question of what counts as copying becomes genuinely hard. If an AI agent trained on public code produces something that closely mirrors an existing open source project, who is responsible? The 'agent trace' comment is half-joking but points at a real forensic challenge.

This will become a recurring conflict. Open source maintainers are already frustrated about their work being used to train models. Now the output of those models may directly compete with the products the training data came from.