UK’s Watchdog Shares AI Principles to Ensure Consumer Protection

The UK’s regulatory watchdog, the Competition and Markets Authority (CMA), unveiled its set of seven guiding principles that are aimed at ensuring the responsible development and use of foundation models (FMs) with artificial intelligence (AI). 

The CMA, which is the country’s primary anti-trust regulator, balances the need to foster open innovation and strong competition with prioritizing consumer protection. The September 18 report emphasizes accountability, transparency, and competition within the current AI boom that takes into consideration how versatile AI systems like FMs are capable of adapting to a variety of specific purposes. 

The Seven Principles

Throughout the report, the CMA cautions against the potential harms in the event competition is weak or consumer protection laws are overlooked – including the spread of false information and AI-enabled fraud (or machine hallucinations).

As it concerns big tech, including OpenAI, Apple, Google, and Microsoft, the CMA addresses the strong possibility of market power being consolidated to their respective platforms, which presents the risk of other AI contenders coming into the arena. 

These seven principles, as outlined in the report, speak to the development and use of FMs for a more ethical and regulated AI framework. 

In creating these principles, the CMA engaged with over 70 stakeholders, gathering information directly from leading FM developers such as Google, Meta, OpenAI, Microsoft, NVIDIA, and Anthropic, as well as governments, academics, and other regulators. The CMA also considered publicly available information, including the latest AI research. 

1. Accountability: FM developers and deployers are accountable for the outputs provided to consumers.

2. Access: Ensuring ongoing access to key inputs without unnecessary restrictions.

3. Diversity: Promoting sustained diversity in business models, including both open and closed approaches.

4. Choice: Sufficient choice for businesses so they can decide how to use FMs. 

5. Flexibility: Allowing flexibility to switch and/or use multiple FMs as needed. 

6. Fair Dealing: Prohibiting anti-competitive conduct, including self-preferencing, tying, or bundling. 

7. Transparency: Providing consumers and businesses with information about the risks and limitations of FM-generated content. 

The next update on the CMA’s progress, including the adoption of these seven principles, is expected to be published in early 2024. 

“The CMA’s role is to help shape these markets in ways that foster strong competition and effective consumer protection, delivering the best outcomes for people and businesses across the UK. In rapidly developing markets like these, it’s critical we put ourselves at the forefront of that thinking, rather than waiting for problems to emerge and only then stepping in with corrective measures,” said Sarah Cardell, CEO of the CMA.

In the report, Cardell stressed the need to proactively address potential issues and not wait for problems to emerge before taking corrective measures. 

“The goal is to further develop these principles collaboratively and ensure that the responsible development of FMs promotes competition and protects consumers.”

Where’s the U.S. At?

In the U.S., governments at various levels are still figuring out how to use and regulate AI-powered tools like ChatGPT. 

Back in June, U.S. Senate Majority Leader Chuck Schumer (D-NY) predicted that new AI legislation was just months away from its final stage, which appears to be on the same trajectory as the EU and its final stages of negotiations for its EU AI Act. 

For the remainder of 2023 and into the beginning of 2024, city and state agencies are expected to start releasing the first wave of generative AI policies that address these AI-powered tools like ChatGPT. In summer 2024, the Biden Administration is expecting to release its initial guidelines for how the federal government can use AI. 

Editor’s note: This article was written by an nft now staff member in collaboration with OpenAI’s GPT-3.5.