21st Century Competition Authorities Adapt to New Economy

Introduction

Good morning, and thank you for your kind introduction.

It was a great honour to become Chair of the UK's Competition and Markets Authority almost exactly a year ago, and it is a role I am passionate about. Why do I say that? Well, over the 30 years leading up to this post, it's been my privilege to help businesses of all shapes and sizes - from long-established multinationals to scrappy start-ups - to compete, to innovate, to grow, and to create value. And if that experience has taught me anything, it's the phenomenal power of competition. I've seen first-hand, time and again, how competitive rivalry unlocks the creative potential in businesses and the people working there. The result is an immensely powerful, positive force. Positive for people, who get greater choice and quality at more affordable prices. Positive for businesses, which are free - and motivated - to innovate, grow and reap the rewards. And positive for the economy, which benefits from higher productivity, investment, job creation and real wage growth.

So, I have been surprised to hear suggestions that competition authorities are becoming too interventionist. Even, from some, that we are anti-business. As someone with a deep business mindset, I think it's important to correct this suggestion. And in doing so, it's timely to revisit and be really clear about the fundamental principles driving the purpose and actions of agencies like the CMA.

A changing environment requiring competition authorities to evolve and adapt

First, I'd like to step back and highlight some of the fundamental shifts underway in the broader external environment, and how the CMA and other competition authorities are adapting to these. I will then focus in on digital markets and then more specifically on AI.

Let me touch on 3 broad forces. The first of these is The rising cost-of-living and deteriorating trust in business. And I think there is an important link between them. Tough economic times are precisely when we need markets to work well, with thriving competition bringing people affordable choices and people being treated fairly. That means competition authorities doubling down on their responsibilities, particularly in the areas that matter most to people - having somewhere to live, being able to travel, to feed and look after ourselves and others. I'm proud of the CMA's recent work in all these areas, most recently around transparency of road fuel prices, for example.

But a persisting cost-of-living crisis against a backdrop of rising business profits in certain sectors can cause some people to question what's happening here and whether it's right. So, where competition authorities find businesses breaking the law with cartel behaviour, or exploiting dominant positions to limit choice or raise prices, it is important to act quickly and unambiguously. Whilst most businesses play by the rules, any wrongdoing by a few during straitened times risks damaging consumer trust and igniting broader anti-business sentiment in a way none of us wishes to see.

So, we've been cracking down on cartels in sectors like construction; as well as taking action to protect household and taxpayer budgets against, for example, anti-competitive wage fixing and excessive pricing of medicines.

Secondly we are also facing far-reaching macro-economic and geopolitical challenges. Economies around the world, with the UK no exception, are striving for growth against an increasingly turbulent backdrop: stagnating productivity, pervasive debt; fragmenting global supply chains; military conflict; a deteriorating climate. And the global consensus needed to tackle these challenges is increasingly fragmented. A 'polycrisis', is how it has been described.

It's understandable and right during such a time that public institutions are called on to play their part in supporting badly needed growth and productivity. But how can we, as competition authorities, do this most effectively? That brings me back to our fundamental purpose: using our powers to create, and safeguard, open and competitive market conditions, level playing fields, and consumer trust in commerce. Taking action where individual businesses seek to undermine that. Ensuring that the innovators and disruptors can continue to push boundaries and expand choice without constraints; that there are no competitive barriers, today or in the future, for the creative thinkers striving for better solutions; for the problem solvers, single-mindedly focused on ever better choice, quality, efficiency and value for the customer.

These are strong drivers of productive and sustainable growth. Arguably, the strongest. As competition authorities, we should be making that link clearer, day-in-day-out, between competitive rivalry, innovation, productivity, and growth. The CMA has a clear and deep sense of our role here. It's central to our strategy. To what we're about and what we do.

The third major force is the ever-accelerating technology revolution and the multitude of new, often disruptive business models this brings and which are transforming how we live and work. What should that mean for competition authorities? Despite all the disruption and novelty, the core questions we should be asking have not (and should not) change, namely: What is the impact on customer choice and alternatives? What about the basis for competition and sources of competitive advantage? How is that advantage created and sustained? Will the power of that advantage eventually yield greater choice, greater innovation, greater value for customers, or less? Will it allow, spur, creative rivalry, or freeze out rivals, freeze out new entrants and future innovators?

The questions to ask are not changing, but the answers are. Nowhere more so than digital markets, which have taken us into very different territory than the brick-and-mortar business models that characterised industry in the past. There's been much discussion in recent years about what makes these markets so challenging from a competition perspective. Massive fixed costs and near zero distribution costs; unprecedented network effects, at a global scale thanks to the universality of the customer need and proposition; leading to extremely high scalability and platform dependencies, and the resulting winner-take-most dynamics. In addition the strategic advantage of a data empire in a world where these are the key raw materials needed to compete - particularly where that data have value in more than 1 market, meaning market power can quickly bleed across into multiple others. And, finally, the rapid pace of scaling and of technology development mean that competition harms can quickly spread and become locked in.

Competition authorities, including the CMA, are consciously and explicitly adapting to this third major force, and I'll explain a bit about what that looks like in practice for the CMA.

Firstly, merger control: Time and again, we've seen that the most effective path to open markets and healthy competition is preventing dominance from developing in the first place. By protecting competition, effective merger control plays a crucial role in driving innovation, productivity, investment and growth. And it is, by definition, forward-looking, looking to pre-empt and prevent a significant reduction in competition. And clearly, a forward looking- analysis is particularly important in dynamic and rapidly evolving markets.

Does that create a presumption or 'ideology' that all mergers should be blocked? Of course not. The facts speak for themselves. The vast majority of deals proceed under the UK system.

But let me be clear: where already dominant companies argue for further market consolidation, they are effectively arguing for weaker competition. For an outcome which will impede markets operating freely. Even in some cases, it might be suggested, for monopolies or oligopolies free from interference. And to do so in the name of innovation or growth is to present an entirely false, and disingenuous, choice between this and the proven benefits of competition.

Complementing merger control, the second lever is ex-ante regulation of those digital markets in which power is already substantial and entrenched. Was entrenched market power something of an inevitability given the fundamental economics at play, or would a different approach to merger control have prevented it? Hindsight is always 20/20, as they say. What is clear today is we must address the effects and the root causes of entrenched market power in these markets, many of which are so central in all our lives. Systemic, even. So, what are the options now? Post-hoc interventions are important but tend to be complex and adversarial in nature. We see a more pragmatic, pro-active, and participatory path forward, in the form of a new legislation - the Digital Markets, Competition and Consumers Bill, which we hope will become law next year.

The Bill has been specifically designed to address challenges that have dogged traditional competition law approaches to digital markets, while avoiding overly burdensome, rigid regulations which could impede innovation and growth. These are fast-moving, cutting-edge sectors, brimming with diverse activities and business models, all prone to disruption. No set of perpetual, one-size-fits-all rules will work here. The principles and tools in the Bill make sense for exactly these reasons. I'll outline them for you, briefly:

• firstly, a more bespoke approach. Standard principles of healthy competition apply, of course - open choices, fair trading, trust, and transparency. But designating firms with Strategic Market Status (meaning substantial and entrenched market power) allows conduct requirements to be much more targeted and tailored to fit different businesses and activities

• secondly, Strategic Market Status designations will be time-bound for review every 5 years, to reflect market and technology shifts which may bring material changes from a competition perspective

• finally, and this is critical - the foundation of the new regime is ongoing, active, and constructive dialogue with potential designated firms and broader stakeholders. This rolling engagement programme will help us sustain our deep understanding of markets and technologies to inform our assessments and any potential interventions. That way, we can work to shape pro-competitive outcomes with a constant eye to the future. Our work in search, social media and digital advertising, mobile ecosystems and indeed foundation models (which I will come to next) are all good examples of this

This is a very different approach to regulating digital markets and I'm pleased to say the programme of stakeholder and industry engagement is already underway and hugely productive.

AI and Foundation Models

Now, the next and arguably greatest, wave of technological revolution we've yet encountered is artificial intelligence. It's no longer new but it is currently undergoing something of a curve jump thanks to a new generation of extremely powerful 'foundation models', which hold the promise of transformational benefits for the way we live, learn, interact and work.

Given the speed of developments, and in the spirit of being more nimble and forward-looking, we acted promptly with an initial review of these models. The goal was not to tackle every issue - important conversations are underway around privacy, security, and intellectual property for example, which are outside of our remit. What we have tried to do is develop an early understanding of the opportunities and risks for competition and consumer protection. Of what it will take to guide this pioneering market, developing at such pace, toward the most favourable outcomes - and what the CMA's role should be in helping to make that happen.

To do that, we explored a broad spectrum of possible scenarios for the development of FM markets and their uses - some with positive outcomes from a competition and consumer protection standpoint, others with outcomes that would cause concern. Our analysis is based on engagement with many stakeholders - from developers of foundation models and their customers, to consumer and industry groups. And of course, a thorough review of publicly available information and the latest AI research.

We can't predict the future, but we can think through the key drivers for each scenario. This led us to the crux of the work: 7, proposed, competition and consumer protection principles to guide market development towards the positive outcomes we all wish to see for people, businesses, and the economy. As with the new digital markets regime, we firmly believe achieving these outcomes should be a collective endeavour. So we approached this work very much in listening mode, with an open door, and I'm delighted with the engagement we received.

I'll give you a whistlestop tour of our findings and preface it by emphasising that our goal for the next stage of work is an even broader programme of engagement to refine the proposed principles and start building a consensus around how we can best implement them to drive those positive outcomes. We welcome the participation and partnership of all of you in this work.

Coming to the review itself, then. We looked at 3 levels of the value chain: competition in foundation model development; the impact of foundation models on competition in other markets; and consumer protection. We also looked at an overarching component, which is the accountability of developers and deployers of these models.

Starting with competition in development

Around 160 foundation models - open and closed source - have been released since 2018, with each new iteration bringing enhanced capabilities and performance. Today's market includes a wide range of firms - some familiar names, like Google, Meta and Microsoft. But also new companies, like Stability AI, Anthropic, and OpenAI whose ChatGPT model catapulted these systems into the media and public discourse.

Sustained, effective competition between developers creates the incentive to continue developing high-quality models, to continue to strive for improvements and breakthroughs. But the strength of competitive tension going forward will likely depend on whether firms can access the resources required for development - the fundamental building blocks of capital, technical expertise, and access to very large, very costly amounts of computing power, and data. These last 2, especially. The more access is restricted, the more likely that only a handful of the biggest, best-resourced firms will build and own leading models. These surviving firms would have the ability and the incentive to impose unfair prices and terms, and to restrict new entrants by developing only closed-source models. The current link between model size and performance creates just such a risk of market consolidation around a few players powerful enough to access key inputs. And all of this could be exacerbated if powerful firms are able to use investments, partnerships, vertical integration, or dominant positions in other key markets to restrict competition and access to this one.

A positive outcome, then, would be an environment where early, successful developers face strong and sustained competitive constraints that counterbalance these risks. A thriving ecosystem of multiple developers, independent of each other, with different business models, able to enter and compete in the market on an ongoing basis. So how do we achieve that?

We propose 2 principles here:

• firstly, access: By which we mean ongoing, ready access to key inputs - data, computing power - on fair terms, without undue restrictions. Related to this the continuing possibility for effective challenge by new entrants to ensure those advantages don't become entrenched and disproportionate positions of power. And ensuring powerful partnerships and vertical integration don't stymie that competitive effect

• secondly, diversity: meaning sustained diversity of business models. Not just closed source, but also open models which could help bring down or keep down barriers to entry and expansion

Next, the impact of foundation models on competition in other markets

It's early days. Many companies are still considering possible use cases for foundation models. But it's already clear they could become deeply embedded in business, the economy, and our daily lives. Improvements to existing products and services, as well as potentially novel uses, are appearing almost daily. From content creation and coding, to productivity software, scientific discovery and healthcare. The potential catalytic effect is certainly exciting, but what would enable this and what could get in the way?

Competition in these downstream markets is likely to be affected by the range of options that customers - be that businesses or consumers - have for choosing, using, and switching between foundation models and associated services. A negative outcome would be 1 in which powerful firms can restrict switching, locking customers into ecosystems with a limited range of deployment options, products, and services. Similarly, if developers were to use their market power (including vertical integration and partnerships) to restrict rivals' access to adjacent or downstream markets for products and services.

A positive scenario, then, would mean customer access to a range of models, and solutions using these models, to suit their needs. It would mean the flexibility to switch easily between them without friction or lock-in, or to use several at once. And those choices would be meaningful, not encumbered by unfair restrictions or anti-competitive conduct based on market power in this, or another market.

The 3 principles we're proposing here are:

• firstly, choice: sufficient options for businesses in how they decide to use models, be that developing in-house, or accessing them via partnerships, APIs, or plug-ins

• secondly, flexibility: Genuine flexibility to switch, or use multiple models or environments according to need, which would require interoperability and ease of porting data

• and thirdly, fair dealing: customers should have confidence that the best products and services will win out, meaning a fair market, free from anti-competitive conduct including self-preferencing, tying, or bundling to entrench or leverage market power

Consumer protection

I want to shift perspective now to how we as people will interact with, and be affected by, these new models. The potential is there to help us all become more productive. To learn, create, imagine, and build. But extensive research has shown that this all depends on the trust users have in these systems. Consumer confidence will depend on people receiving reliable, accurate information and fair treatment from foundation model-based services. And we have already learned from earlier digital markets how important it is to stay vigilant here.

We've all seen that some technical features of foundation models that can make them unreliable, producing false or misleading results, known as 'hallucinations'. And it can be difficult for consumers to know who's accountable and responsible if something does go wrong. More broadly, in many instances, they'll be unaware they're even engaging with a foundation model. This is particularly concerning given the potential we heard from stakeholders for foundation models to engage in deceptive or manipulative conduct.

So there are 2 types of harm to protect against here. First, in the hands of bad actors, new harms are emerging, like deep fakes. Second, the existing harms which consumer protection authorities around the world are already battling - like fake reviews, hidden advertising, and phishing scams - could get much worse. This all mirrors familiar trends in digital markets and relates closely to CMA work on online choice architecture, pressure selling, and drip pricing. We welcome the voluntary commitments made by developers to tackle the risks but clearly there is much complex and careful work still ahead.

Let's focus on the positive scenario now. All firms in this space - model developers and business customers - are expected to comply with consumer protection law. But vigorous competition is also likely to be a powerful antidote to laxity in responsible development and deployment, because it organically places a high value on consumer trust. A positive outcome, then, would be 1 where developers have incentives to compete not just on technical innovation, but on model reliability, accuracy and transparency. In putting the customers' interests first. Incentives like the proper allocation of responsibility and accountability, as well as a mechanism for redress. in this scenario, consumer awareness and understanding is high, and both developers and businesses ensure appropriate safeguards are in place to protect people from harm.

So our final 2 proposed principles are:

• transparency: That consumers and businesses have sufficient, understandable, and accurate information about model uses, risks, and limitations, so they can make informed choices. And developers give businesses deploying foundation models the right information for them to manage their responsibilities to consumers

• and Accountability: Developers and deployers are accountable for outputs provided to customers and take responsibility for ensuring appropriate safeguards are in place to protect them from harms

Conclusion

I'll finish by making a broader point which feels particularly apt in this room today, amongst colleagues and partners at fellow competition authorities, and members of the broader competition community advising consumers and businesses. And that is around the power of dialogue and collaboration.

Our role as competition authorities is not just enforcement. It's to foster an environment where businesses flourish, innovate, and compete fairly. But to achieve that, in our globalised world, with all its global challenges, we need global collaboration. We operate in different jurisdictions, often with different consumer and competitor market contexts, and naturally we have different approaches to some specific issues. But the increasingly concentrated industries we deal with are global, and the harms we are committed to preventing, particularly in this digital age, are borderless. Anti-competitive behaviour, or the erosion of competition in 1 country, resounds around the world in a way it did not half a century, or even a few decades ago.

The value of dialogue and collaboration between us has never been more important. We are all building enhanced capabilities and skills, developing novel solutions to the pressing challenges I have talked about today. We redouble the value of that important work when we share ideas, learnings, and insights. When we engage in a global conversation, as we are doing this week, our collective knowledge and strong working relationships are powerful public goods. Events like this are a testament to that, and I have great faith that we can continue to adapt and work together to keep the spirit of competition alive and well around the world. That work is needed now, more than ever.