How Open-Source Developers Can Help Achieve Democratic Control of Artificial Intelligence through the EU A.I. Act

**The EU’s A.I. Act: Shaping the Future of Artificial Intelligence Regulation**

As the EU prepares to finalize the world’s first comprehensive A.I. regulation, policymakers and open-source software developers are taking note. The A.I. Act will play a crucial role in shaping how artificial intelligence is built, deployed, and regulated globally, following the precedent set by the GDPR in tech regulation. The Act aims to encourage responsible A.I. innovation while mitigating potential risks associated with the technology.

**The Need for Responsible A.I. Development**

The A.I. Act recognizes the need for responsible development, deployment, and use of A.I. systems. Without responsible practices, A.I. can have harmful effects, such as biased algorithmic decisions impacting access to opportunities and the spread of misinformation. Therefore, the Act introduces a risk-based approach to regulating A.I. across all sectors. Before A.I. systems can be sold or deployed, they must meet specific requirements related to risk management, data transparency, documentation, oversight, and quality. The level of requirements may vary depending on the system and use-case. Sensitive areas like critical infrastructure and determining access to education and employment opportunities are considered high-risk.

**Open-Source Contributions and the A.I. Act**

The A.I. Act reflects many responsible A.I. practices that have been pioneered in the open-source community, including the use of model cards and datasheets for datasets. These tools provide valuable documentation of an A.I. system’s capabilities, limitations, and design decisions. Open-source developers have been crucial in driving A.I. innovation, contributing to the development of essential frameworks like PyTorch and TensorFlow and creating plugins that enhance A.I. capabilities.

**Recognition of Open-Source Contributions**

The global developer community has embraced open-source A.I. models, contributing to the sustainability, inclusivity, and transparency of A.I. These models have been optimized to run efficiently on mobile devices, support minority languages, and provide academic researchers access to study their inner workings. The A.I. Act acknowledges these open innovations and recognizes the value they bring to A.I. development. The Parliament text includes a risk-based exemption for open-source developers, ensuring that collaborating and building A.I. components openly is protected. While open-source developers are encouraged to follow documentation best practices, compliance with regulations is the responsibility of the entities that incorporate open-source components in A.I. applications.

**Improving the A.I. Act**

While the A.I. Act demonstrates positive steps towards regulating A.I., there is room for further improvements. As the Act aims to address the rise of generative A.I., provisions for providers of foundation models have been included. However, it remains unclear how open-source developers, academics, and non-profits will be able to comply with obligations designed for commercial products. In the final negotiations, it is crucial to consider these issues and ensure that any requirements for foundation models are risk-calibrated and feasible for implementation.

**Global Reach of the A.I. Act**

The regulations established by the A.I. Act may extend well beyond the EU’s single market. Countries worldwide, including Canada, Brazil, the United States, and China, have proposed or are considering A.I. regulation. The recognition of open-source innovation’s value is growing, as it contributes significantly to GDP and will continue to do so with A.I. advancements. The EU’s experience in tailoring regulations to support open-source developers can serve as an important example for others to learn from. Countries like the U.S. are also engaging in comprehensive A.I. regulation discussions and should include open-source developers in these conversations.

**A.I. Regulation in Canada**

Canadian policymakers have been diligently working on legislation to create an A.I. regulator that follows a risk-based approach similar to the EU. This legislation aims to protect developers involved in collaborative A.I. development while ensuring responsible practices.

**A Unified Approach to Responsible A.I. Development**

As A.I. technology and policy rapidly evolve worldwide, a risk-based approach remains a common thread in responsible A.I. development. The EU’s A.I. Act and similar initiatives recognize the importance of encouraging innovation while safeguarding against potential risks associated with A.I. By setting clear regulations and acknowledging the value of open-source contributions, policymakers can shape the future of A.I. in a responsible and inclusive manner.

Note: The opinions expressed in this article are those of the author and do not necessarily reflect the opinions or beliefs of Fortune.