UK Antitrust Watchdog Launches Review of Generative AI Foundational Models

UK's Antitrust Watchdog Initiates Review of Generative AI

The UK's Competition and Markets Authority (CMA) has launched an initial review into "AI foundational models," the core technology behind generative AI tools like OpenAI's ChatGPT and Microsoft's New Bing. This move signals a proactive approach by regulators to understand and shape the rapidly evolving AI landscape.

Key Focus Areas of the CMA's Review:

The CMA's review aims to assess the competitive dynamics and consumer protection implications associated with AI foundational models. The primary objectives include:

• Understanding Development and Use: Examining how these models are being developed and deployed.
• Assessing Competition: Identifying potential competition concerns and opportunities in the market for foundational models and their applications.
• Consumer Protection: Evaluating risks related to consumer protection, such as misleading information or unfair practices.
• Guiding Principles: Developing principles to foster responsible innovation and ensure fair markets.

Timeline and Stakeholder Engagement:

The CMA plans to publish its initial findings in early September. To inform its work, the authority has set a deadline of June 2 for stakeholders to submit their responses and insights.

Government Mandate and Regulatory Principles:

This review is conducted under a government directive, emphasizing five key principles for AI development and deployment:

1. Safety, Security, and Robustness: Ensuring AI systems are reliable and secure.
2. Appropriate Transparency and Explainability: Making AI decision-making processes understandable.
3. Fairness: Preventing bias and discrimination in AI applications.
4. Accountability and Governance: Establishing clear lines of responsibility for AI systems.
5. Contestability and Redress: Ensuring open markets and mechanisms for addressing harm.

Defining Foundational Models:

The term "foundational models" was coined by Stanford University's Human-Centered Artificial Intelligence Center. It refers to AI systems trained on vast datasets, which can then be adapted for a wide range of applications.

CMA's Mandate and Future Powers:

The CMA's core mission is to support open and competitive markets. The review aligns with this mission and anticipates potential future powers for the CMA to regulate Big Tech, including AI providers, under the Digital Markets Unit's framework.

Broader Regulatory Landscape:

• UK ICO: The UK's Information Commissioner's Office (ICO) is also focusing on generative AI, emphasizing data protection by design and default for developers and users.
• European Union: EU lawmakers are developing a comprehensive AI rulebook, with ongoing negotiations to regulate foundational models within a risk-based framework. Concerns include safety, supply chain responsibilities, and copyright issues.
• Privacy Investigations: Several EU data protection authorities, including Italy's and Spain's, are investigating AI chatbots like ChatGPT for potential GDPR violations.

Related Content:

• UK to avoid fixed rules for AI – in favor of ‘context-specific guidance’
• Microsoft doubles down on AI with new Bing features
• EU lawmakers eye tiered approach to regulating generative AI

Image:

Image Credits: Leon Neal / Getty Images

Topics:

• AI
• Europe
• Government & Policy
• Privacy
• United Kingdom