Introduction
As the use of Artificial Intelligence (AI) rapidly increases, it has brought unprecedented innovation and efficiency across various sectors. Competition regulators face a real challenge of ensuring that there are sufficient protections in place to guard against unintended potentially anti-competitive market effects. As AI continues to evolve, regulators and legal professionals are grappling with its implications on fair competition, market dominance, and consumer welfare.
The purpose of competition law is to maintain market competition by preventing anti-competitive behaviour and to promote consumer welfare, innovation, and a level playing field for businesses. In the EU and UK, the primary legal frameworks in the EU and UK are:
- Article 101 of the Treaty on the Functioning of the European Union (TFEU) and Chapter 1 of the Competition Act 1998 (CA98): Prohibition on agreements that restrict competition, for example, price fixing or market sharing cartels.
- Article 102 TFEU and Chapter 2 CA98: Prohibition on the abuse of a dominant market position.
- Merger regulation: Controls mergers and acquisitions to prevent excessive market concentration.
There is a delicate balance between encouraging innovation and maintaining fair competition. As was found to be the case with the rapid expansion of big tech in the last few years, existing competition enforcement powers may not be sufficient to protect competition and markets from the impact of transformative technology. Here, we explore how to navigate an AI future, based on lessons from the past.
How it began – Pricing algorithms
A pricing algorithm is a set of computer rules and processes designed to optimise the price for a product or service. These algorithms analyse various factors such as market conditions, competitor prices, demand fluctuations, and customer behaviour to automatically adjust prices. The use of such algorithms is not illegal; however there has been a concern that they may facilitate both explicit and tacit collusion between competitors in breach of competition law.
Explicit Collusion
In 2016, the UK's Competition and Markets Authority (CMA) found that two online poster sellers agreed that they would not undercut each other’s prices for licensed sport and entertainment posters and frames that were sold on Amazon’s UK website (Case 50223: Online sales of posters and frames. Trod Ltd v GB Eye Ltd [2016]). The agreement was implemented using automated repricing software which the parties each configured to give effect to the agreement.
The software enabled sellers to compete with other online sellers by automatically adjusting the prices of their products in response to the live prices of competitors’ products. The adjustments were based on the settings determined by each seller and once in place, the software worked to adjust the seller's prices automatically (every 15 minutes) in response to competitors' pricing. The software allowed certain competitors to be excluded from the algorithms if they were added to the ‘ignore list’. The two sellers in this case had added each other to the ignore list, which ensured that they would never undercut each other in line with their agreement.
While the parties to this case did not themselves actively monitor the pricing on Amazon, they enabled their software to limit price competition and form the basis of an illegal pricing cartel. This was a novel basis for a Chapter 1 decision at that time, and only came to light as a result of one of the parties "blowing the whistle".
For its own part, the Court of Justice of the European Union (CJEU) ruled in 2016 on a preliminary reference from Lithuania, in relation to an alleged concerted practice in the packaged tours sales market (Case C-74/14 UAB Eturas and Others v Lietuvos Respublikos konkurencijos taryba). The conduct took place within the E-turas system in Lithuania, an online computer reservation system for the search and booking of package tours, which sent messages to travel agents through the online system confirming its pricing algorithm would cap discounts at 3%. Any discounts above that would automatically be reduced in line with the policy. The CJEU confirmed that this was a breach of competition law. The decision deals at length with the burden of proof and what agents should have done to distance themselves but what is relevant here is that, through a platform acting without human input and used by all travel agents, a breach of the law was found.
Tacit collusion
Regulators have expressed concern that the use of pricing algorithms could lead to tacit collusion between entities (CMA94, 8 October 2018: Pricing Algorithms: Economic working paper on the use of algorithms to facilitate collusion and personalised pricing). An increased use of pricing algorithms will result in faster pricing shifts, increased market transparency, and a perfect storm for collusion:
- Hub and spoke cartel: where multiple market players use the same pricing algorithm, they may react similarly to market dynamics, and end up with similar pricing patterns. This may lead to an indirect exchange of information and coordination of pricing policies if users are aware that the algorithms are used by their competitors.
- Predictable agent: Simple pricing algorithms that react predictably to market conditions, such as 'lowest price matching,' can lead to transparent and predictable pricing, resulting in tacit collusion and parallel pricing.
- Autonomous algorithms: Sophisticated algorithms can learn independently and tacitly coordinate prices. If the algorithm is instructed to maximize profit, it might determine through trial and error that aligning prices with competitors is the most profitable strategy. Big data plays a crucial role in these situations as the more data available to an algorithm, the better the results. We have considered this further below.
With the above in mind, the use of algorithms to help execute a cartel’s work is in effect the same as a cartel executed by individuals: technology facilitates anticompetitive behaviour that humans would otherwise have carried out. In the words of the European Commissioner for Competition Margrethe Vestager: "companies can’t escape responsibility by hiding behind a computer program…what businesses can – and must – do is to ensure antitrust compliance by design".
Where it's going – Artificial intelligence
As noted above "autonomous algorithms" are sophisticated and self-learn software to anticipate market events, potentially leading to "algorithmic collusion". Unlike traditional collusion, where businesses explicitly agree to fix prices, algorithmic collusion can occur independently through AI systems. These systems can analyse market conditions and adjust prices in real-time, potentially leading to collusion without any direct human intervention. It is very difficult to detect and prove algorithmic collusion, and traditional investigative tools might be inadequate, necessitating advanced monitoring technologies and new legal frameworks that can address the subtleties of AI-driven collusion. Regulators are actively researching and developing strategies to identify and mitigate the risks:
Market power and data dominance
AI relies heavily on data, making it a critical asset for businesses. Companies with access to vast amounts of data can develop superior AI capabilities, potentially leading to market dominance. This data-driven advantage can create significant barriers to entry for smaller competitors and startups, reducing market competition. Regulators have already taken steps to address data dominance. For instance, in the Google Shopping case, the European Commission fined Google for abusing its dominant position by using software favouring its own shopping service over competitors. In the UK, the CMA accepted commitments from Amazon where it was alleged, amongst other things, that the entity's algorithms were self-preferencing.
Discriminatory practices and bias
AI systems can inadvertently or deliberately engage in discriminatory practices, affecting consumers and competitors. For example, personalised pricing algorithms can lead to price discrimination, where different consumers are charged varying prices for the same product based on their personal data and purchasing behaviour (it is worth noting that in 2012 and 2017, the UK regulator found no evidence of pricing being different or personalised for different consumers, although there were examples of different consumers being shown different search results on retail websites, including different numbers of results or a different order of results). Similarly, AI-driven ad placements or search engine rankings can unfairly favour certain businesses.
Mergers and acquisitions in the AI era
AI also complicates the landscape of mergers and acquisitions (M&A). Traditional M&A assessments may not fully capture the competitive impact of AI technologies. When companies with significant AI capabilities or large datasets merge, the potential for increased market concentration and reduced innovation is significant.
Regulatory adaptation and innovation
The rapid pace of AI development poses a challenge for the relatively slower-moving regulatory frameworks. There is a need for adaptive and forward-thinking regulatory approaches to keep pace with technological advancements. This includes continuous monitoring of AI developments and proactive engagement with technology experts and industry stakeholders. To address the dynamic nature of AI, regulators are exploring the adoption of principle-based regulatory frameworks, which should allow for quick adjustments in response to new AI capabilities and market behaviours.
They are also trying to better understand the technology. Many authorities are recruiting data computer specialist or doing studies in the area (such as the CMA initial review of foundation models).
Regulatory Initiatives
The European Union (EU) and UK have already launched several initiatives to address the intersection of AI and competition law:
- The Digital Markets Act (DMA): this EU legislation aims to ensure fair and open digital markets by regulating large online platforms, or "gatekeepers," that could potentially abuse their market power. It is unclear if AI is covered only where the gatekeeper has integrated it within the gatekeeper designated services.
- The AI Act of the European Union: this is the AI Act is the world's first comprehensive legal framework for regulating artificial intelligence. It will have far-reaching implications for both the development and use of AI; however, the practical consequences for providers and users are still unclear in many areas. In relation to competition law, the impact will be on competition enforcement, for example, the broad powers of procedure provided to the relevant supervisory agencies, which include examining evidence and accessing data and documents, are transferrable to national competition authorities.
- The Digital Markets Competition and Consumers Act 2024 (DMCC): the DMCC is due to come into force in Autumn 2024 and is the UK equivalent of the DMA. It will give CMA the ability to respond quickly and flexibly to the often-rapid developments in digital markets, including through setting targeted conduct requirements on firms found to have strategic market status (SMS) in respect of a digital activity. The CMA has confirmed that AI and its deployment by firms will be relevant to the CMA’s selection of SMS candidates, particularly where AI is deployed in connection with other, more established activities. Further, the CMA will be able to understand the effects of a designated firm’s algorithms on competition and consumers.
- Merger Control - Article 22 and DMCC: both the EU and UK have now power to review transactions that do not meet merger thresholds and this may be relevant to AI. Both regulators have recently considered whether the partnership between Microsoft Corporation and OpenAI Inc fell within the scope of its merger control and the CMA is also considering whether the Microsoft has "effectively" merged with Inflection AI, a tech start up, as a result of Microsoft’s hiring of certain former employees of Inflection and its “associated arrangements” with the company. In addition, the DMA and DMCC both place reporting obligations on gatekeepers and SMS firms, in the event of proposed acquisitions, which may lead to investigations where there are no obvious competition concerns or overlaps.
Conclusion
AI is undeniably transforming the economic landscape, offering significant benefits but also posing complex challenges for competition law. Regulators are addressing these challenges, developing new regulatory tools and frameworks to ensure that AI-driven markets remain competitive and fair. As AI continues to evolve, it is clear that regulators have learned from their past mistakes and adopting a proactive approach in adapting competition law will be crucial in navigating the future. By fostering innovation while safeguarding consumer welfare and market fairness, competition regulators will ensure the responsible and competitive deployment of AI technologies.
How we can help
Our competition team is assisting clients in competition issues in this new area. If you would like to discuss it please contact Paolo Palmigiano or Louisa Penny.
