Preventing the concentration of power in artificial intelligence

Together with European partners, we have called on the UK Competition and Markets Authority (CMA) to further investigate Microsoft’s involvement in OpenAI. It is an example of the growing dominance of big tech companies in the AI sector. The concentration of economic power threatens to be exacerbated by the AI boom. The antitrust authorities must therefore take countermeasures.

Microsoft’s investments in OpenAI and the turmoil at OpenAI in November 2023 show the influence Microsoft has on OpenAI. This influence is problematic and raises competitive issues. Microsoft plays a relevant role in the operation and strategic direction of OpenAI, which has profound implications for competition between the two companies and for AI in general.

In our statement (pdf) we list a number of issues that should be investigated by the CMA. The UK competition authority has launched an investigation into the Microsoft/OpenAI link and has invited comments. The European Commission has also invited comments on AI.

Microsoft’s investment in OpenAI should be seen in the context of the past two decades of economic history. Weak antitrust enforcement has contributed to a handful of dominant tech companies controlling digital markets – the foundation of the modern economy. These so-called gatekeepers (they decide who gets in through the gate and who doesn’t) use their unprecedented access to computing infrastructure, data and expertise to influence the development and commercialization of AI.

In a fairer and more open digital economy, AI development would be pluralistic and decentralized. It would be more responsive to the needs of consumers, businesses, governments and citizens. Companies like OpenAI or Anthropic could, in time, threaten giants like Amazon, Google and Microsoft.

Unfortunately, this is not the path we are on. Instead, smaller AI companies are becoming increasingly dependent on the infrastructure and financial support of the tech giants. Microsoft’s partnership with OpenAI and the joint 6 billion dollar investment by Google and Amazon in Anthropic illustrate this. The German company SAP is also investing in Anthropic. Allowing the largest incumbents to dictate how AI develops by extending their monopoly power to future markets and technologies serves their profit margins, but not the public interest.

If Big Tech can extend their current monopoly power to AI without opposition, they will dictate the development of AI. This will serve their profit interests, but not the public interest. That’s why regulators need to take early and aggressive action against anti-competitive behavior in the AI space.

Artificial intelligence is often seen as something completely new. Something that poses new challenges and requires new laws and rules. But what if we changed the way we look at AI? A study by the Open Market Institute looks at AI through the lens of the power and behavior of the corporations that control the basic building blocks of AI. The authors found that Google, Amazon and Microsoft control more than two-thirds of the global cloud market. This gives them access to computing capacity that no other player has. The companies also control large data sets for training AI models. Monopolistic structures are also emerging for the computer chips required for AI.

A handful of large technology companies could control the future of artificial intelligence by exploiting their monopoly power and aggressively capturing other players. The monopolization of AI would exacerbate existing problems of digitalization, such as

• the creation and dissemination of misinformation,
• the decline in news and high-quality journalism,
• the increase in personalized advertising and online addiction,
• the exploitation of workers,
• the monopolistic abuse and exclusion of smaller companies and competitors.

To counteract this development, the study makes far-reaching demands to combat monopoly power in AI:

• Competition authorities should prevent dominant tech companies from monopolizing the AI market through acquisitions, investments and partnerships. Existing mergers and partnerships such as Google’s acquisition of DeepMind in 2014, Microsoft’s significant stake in OpenAI and Amazon’s recent partnership with Anthropic should be scrutinized.
• Any discrimination by Microsoft, Google & Co in the provision of basic AI services for private individuals and companies should be prohibited.
• Cloud computing should be recognized as critical digital infrastructure that should be separate from the large gatekeeper platforms; it should be regulated like public utilities.
• Copyrights should be consistently enforced to protect the property of authors, artists and independent publishers from misappropriation and abuse by tech companies.

Artificial intelligence certainly presents our societies with new challenges. But new rules are not needed for everything. For the major problem of the concentration of power, an important step would be to apply the existing competition law consistently. The debate on this is in full swing. It is now important for the competition authorities to take concrete measures. As civil society, we should work towards this.