The UK’s Competition and Markets Authority (CMA) has outlined a set of specific principles to guide AI regulations and companies involved in its development. The CMA, which serves as the primary antitrust regulator for the UK, focused its attention on foundation models – AI systems like OpenAI’s GPT-4 and Meta’s Llama 2 – which form the basis for many generative AI use cases.
According to the CMA, companies that create foundation models should adhere to seven principles. These principles include ensuring that developers and businesses using these models are accountable for the output provided to consumers, ensuring broad access to the necessary chips, processors, and training data for developing AI systems, and offering a variety of business models, including both open and closed models. Additionally, companies should allow businesses to choose how to use the model, provide flexibility to switch to other models or use multiple models simultaneously, avoid anti-competitive actions such as bundling or self-preferencing, and offer transparency regarding the risks and limitations of generative AI content.
The CMA developed these principles after conducting an initial review and plans to engage in a series of dialogues with consumer and civil society groups, foundation model developers (including Google, Meta, OpenAI, Microsoft, Nvidia, and Anthropic), foundation model users, and academics.
The agency believes that providing principles for the development and deployment of foundation models is necessary to protect competition and prevent the proliferation of low-performing AI systems. It believes that vibrant competition and innovation could benefit the economy as a whole through increased productivity and economic growth. On the other hand, if competition is weak, “a handful of firms gain or entrench positions of market power and fail to offer the best products and services or charge high prices.”
While AI regulation raises broader questions regarding copyright and data privacy and protection, the CMA chose to focus on competition and consumer protection to guide the current development of technology.
Notably, governments worldwide have been exploring different ways to regulate generative AI. For example, the European Union’s proposed AI Act also focuses on foundation models and requires companies to comply with transparency rules. China’s recent AI rules mandate AI companies to register with the government and promise not to offer anti-competitive algorithms. In the United States, policymakers are still determining how to approach AI regulation, with some hoping to have rules in place by the end of this year.
In conclusion, the CMA’s specific principles for AI regulations aim to ensure accountability, access, diversity of business models, choice, flexibility, non-anti-competitive behavior, and transparency in the development and deployment of foundation models. By focusing on competition and consumer protection, the CMA aims to foster vibrant competition, drive innovation, and prevent the consolidation of market power. These principles align with ongoing efforts worldwide to regulate AI and address various challenges in the field.