The EU A.I. Act seeks democratic artificial intelligence control by involving open-source developers. As policymakers finalise the world’s first comprehensive A.I. regulation, responsible innovation and risk mitigation take centre stage.
The EU’s groundbreaking A.I. Act is set to revolutionise the regulation of artificial intelligence. EU policymakers are in the final stages of implementing the world’s first comprehensive framework for governing artificial intelligence. The forthcoming A.I. Act aims to shape A.I.’s global development, deployment, and regulation following the EU’s influential General Data Protection Regulation (GDPR). As the EU establishes itself as a leader in tech regulation, policymakers worldwide are taking notice, including open-source software developers.
The A.I. Act adopts a risk-based approach, setting requirements for A.I. systems in various sectors. Before deployment or sale, systems must meet criteria related to risk management, data transparency, documentation, oversight, and quality. Sensitive areas such as critical infrastructure and access to life opportunities in education and employment are classified as high-risk. Generative A.I., like ChatGPT, also falls within the Act’s scope, with provisions specifically addressing it. Negotiations among the EU Parliament, Council, and Commission are underway to finalise the law by year-end.
The Act acknowledges the value of responsible A.I. tools pioneered by the open-source community, including model cards and datasheets for datasets. Open-source developers are pivotal in driving A.I. innovation, maintaining essential frameworks like PyTorch and TensorFlow, and creating plugins for systems like ChatGPT.
Recognising the significance of open source innovation, the Parliament’s text includes a risk-based exemption for open-source developers, allowing collaboration and protection for developers incorporating open-source components. Compliance with best practices, such as model and data cards, is encouraged but falls on entities using open-source components. The Act’s provisions offer clarity for developers and should be part of the final law.
However, some aspects of the Act can be improved. Concerns arise regarding obligations for open-source developers, academics, and non-profits, particularly with requirements tailored for commercial products. As negotiations progress, it is crucial to calibrate requirements for foundation models based on risk levels and feasibility.
The impact of the EU’s A.I. regulations may extend beyond the single market, with other countries considering A.I. regulations. The EU’s experience can serve as an example, demonstrating how regulations can support open-source developers. U.S. Senate Majority Leader Schumer has unveiled a bipartisan framework for A.I. regulation, highlighting the need to include open-source developers in discussions.
Canada, too, is actively working on A.I. legislation with a risk-based approach that protects collaborative A.I. development.
As A.I. technology and policy continue to evolve globally, a risk-based approach facilitates responsible collaboration in A.I. development. The EU’s A.I. Act represents a significant step towards democratic artificial intelligence control, with open-source developers playing a crucial role. By recognising and supporting their contributions, the Act can shape responsible A.I. development and deployment in the EU and worldwide.
