Open-Source AI
AI models and tools released with open licenses allowing free use, modification, and distribution.
Overview
Open-source AI encompasses models, datasets, and tools released under permissive licenses that allow the community to use, study, modify, and distribute them. Key examples include Meta's LLaMA, Mistral's models, Stability AI's Stable Diffusion, and frameworks like PyTorch and TensorFlow.
Debate
The definition of 'open source' in AI is debated. Some releases include model weights but not training data or code (open-weight vs truly open-source). The Open Source Initiative proposed the Open Source AI Definition requiring training data, code, and weights. Open-source AI democratizes access but raises concerns about misuse, safety, and the ability to enforce responsible use guidelines.