The EU AI Act aims to support open-source AI development by offering special exemptions. For AI models that are made freely available to the public, like Meta’s LLaMA, the requirement to disclose details about the training data is waived. This provision helps promote transparency and collaboration in the AI community, while reducing regulatory burdens on open-source projects.

Source: EUR-Lex
Many supporters of open culture are excited about the emergence of generative AI. In most countries today, only content created by humans is eligible for copyright protection. This means that using AI-generated content can be done with lower legal risks. Furthermore, when AI generates content, it produces a new, unique expression that differs from the original copyrighted work. Considering this, it is doubtful that content created by AI would violate the copyright of the original piece.
The development of AI technologies relies heavily on an open culture. AI is already beginning to impact many jobs and industries. There are concerns that if a small number of large tech companies gain monopoly control over the most advanced AI models, they will reap all the commercial benefits while locking out competition. This level of concentrated power over such a transformative technology would be problematic.
In contrast, open-source AI models leverage the collective intelligence of the broader community. The results and benefits of open-source AI can be accessed and used by anyone. While open-source AI still involves complex issues that need to be worked out, the open and collaborative nature of the development process is more inclusive and holds the potential to benefit society as a whole.
Although there are still challenges such as copyright and privacy with open models, the concept of open models conveys a sense of shared ownership and accessibility for all. As a result, people will likely be more willing to accept open models compared to closed, proprietary ones controlled by a handful of companies.