The major question facing the US right now is one of finding the best policy tool for regulating AI. Tryfonov/adobestock.com 

Antitrust authorities in the US are targeting the new frontier of artificial intelligence (AI) for potential enforcement action.

Beginning in June, the US Department of Justice (DoJ) signalled that it’s investigating high-end chip maker Nvidia Corp, whose shares have more than doubled in 2024. The Federal Trade Commission (FTC) is meanwhile probing Microsoft Corp and whether it’s gaining unfair advantage from an investment in major sector player OpenAI. Nvidia Corp and Microsoft declined to comment when approached by Global Insight.

The regulatory moves come as Facebook parent Meta releases the largest generative AI model yet, Llama 3.1, after an investment estimated in the hundreds of millions of US dollars. 

‘I'm not sure how great a grasp we have on what seems to be a very complicated stack of distinct activities that are all being roped together, tied up in a bow with this concept of artificial intelligence’, says Daniel G Swanson, past Co-Chair of the IBA Antitrust Section and a partner at Gibson Dunn & Crutcher in Los Angeles. 

‘I am not sure that the government, with all of their excitement and zeal to jump into this area, really knows what they're doing, either’, he says. ‘There's a lot of prejudgment, or at least a lot of suggesting the authorities have some sense something pernicious is about to happen or is happening, whereas it’s not so clear that’s the case’.
 
Antitrust authorities are worried about ‘entrenchment of dominance’ and see ‘AI as like a fire accelerant’, he says. But the major tech companies view AI as both a threat and an opportunity. It’s ‘fluid and dynamic’ and there’s an enormous amount of ‘hype’, Swanson adds.

For now, the US government is falling back on antitrust authorities as its major policy tool in this moment, risking potential over-enforcement.

‘We're in something of a state of global disarray when it comes to addressing the digital sector’, says Diana Moss, Vice President and Director of Competition Policy at the Progressive Policy Institute in Washington, DC. ‘The tech companies are very complex, they’ve posed an enormous challenge for antitrust enforcement, and they got really big, really fast’, Moss says. 

As a result, Moss says the DoJ and FTC are now in ‘catch-up mode’, forced to use anti-monopoly laws ‘to discipline competition’ in the tech sector. She explains that ‘these cases are incredibly hard to bring because the burden of proof to show monopoly power is high’. As a result, antitrust actions take years to litigate and often yield only mixed results.

This US approach is very different from the compliance regimes Europe has deployed in its digital markets and AI legislation. ‘Other countries are asking, “what should we do?” But nobody has really hit the nail on the head, in terms of policy approach to disciplining the market power of the techs’, Moss says.

President Joe Biden issued a ‘whole of government’ executive order in autumn 2023 directing US agencies to establish standards for AI use within their fields. The order sets out laudable goals such as safety and security, privacy protection and fairness. Critics say it falls short of a comprehensive AI governance scheme. Congress has been working on legislation but will probably not do anything meaningful until 2025 or 2026.

Part of the struggle for policymakers is that even defining AI is something of an unfathomable moving target, according to computer scientists. ‘What's different with AI is, this is making computers do things that no human understands how to do’, says David Evans, a professor of computer science at the University of Virginia who conducts research into machine learning. ‘Instead of humans working either individually or as a group to write a program, AI works by training an algorithm’, he explains. 
 
New foundational AI systems such as ChatGPT's large language model are based on hundreds of billions of parameters and trained with everything on the internet and terabytes of material from other sources. Newer models are being trained in ways that are ten or 100 times more efficient. ‘Once you scale it up to the size of models that you're training, and the amount of data that you're using to train them, it is very hard to predict, or understand, or have any constraints on what they might do’, Evans says.

If anything, that opacity is encouraging the DoJ’s Antitrust Division to take a close look at the AI ecosystem. ‘Over and over again, we see that antitrust enforcement in moments of industrial evolution has the opportunity to spur innovation in its wake, opening the door to new competitors, allowing for the development of different business models and new economies’, Kanter says.