British watchdog has 'real concerns' about the staggering love-in between cloud giants and AI upstarts
Billions in investment? Yeeeah, right – looks more like ensuring only select few developers thrive
The UK's competition watchdog sniffed around the AI industry with a bit more interest than usual on Thursday at an antitrust event in the US.
Speaking at the 72nd Antitrust Law Spring Meeting in Washington, DC, Sarah Cardell, CEO of the UK Competition and Markets Authority, discussed "growing concerns" that the web of connected partnerships between AI technology companies may hinder competition.
"I think it’s fair to say that when we started this work, we were curious," said Cardell. "Now, with a deeper understanding and watching developments very closely, we have real concerns."
Last September he CMA issued a report on AI foundation models (FM) – the basis for services like ChatGPT – and their impact on consumers and competition. The report proposed a set of principles for AI model vendors, to ensure accountability, access, diversity, choice, and so on.
Seven months on and the CMA now believes that a handful of dominant technology firms – Google, Amazon, Microsoft, Meta and Apple (GAMMA) – may foreclose the possibility of real competition through partnerships, investments, and agreements.
"These firms often have strong positions in critical inputs for FM development – such as large data sets or AI compute infrastructure at significant scale – and/or key access points or routes to market for FM release and deployment," the CMA said in an update paper detailing its reasoning.
"We are therefore concerned that the largest incumbent technology firms could profoundly shape the development of FM-related markets to the detriment of fair, open and effective competition and ultimately harm businesses and consumers."
The CMA, which delayed Microsoft's acquisition of Activision Blizzard, is already investigating Microsoft's partnership with OpenAI, not to mention the cloud infrastructure market. The watchdog's heightened focus on AI means closer attention to merger reviews and hints at the sorts of issues that will get scrutinized under the UK's Digital Markets, Competition and Consumers Bill.
- H-1B visa fraud alive and well amid efforts to crack down on abuse
- US, Japan announce joint AI research projects funded by Nvidia, Microsoft, and others
- Boffins deem Google DeepMind's material discoveries rather shallow
- US House mulls forcing AI makers to reveal use of copyrighted training data
Elsewhere in Washington, DC, officials from the US Justice Department, the Federal Trade Commission, and European Commission convened for the fourth US-EU Joint Technology Competition Policy Dialogue.
EC executive vice president Margrethe Vestager echoed Cardell's remarks by observing the need to keep an eye on AI and the broader technology market.
"The fast-moving technology sector raises global challenges such as regarding artificial intelligence and cloud computing more broadly," she said. "It is essential to anticipate and address such challenges through close cooperation, leveraging our respective experiences for the benefit of consumers and businesses on both sides of the Atlantic."
The antitrust saber rattling isn't likely to alarm large AI vendors like Microsoft, which already cautions that its AI projects may run afoul of regulators in its investor risk boilerplate.
"Our implementation of AI systems could result in legal liability, regulatory action, brand, reputational, or competitive harm, or other adverse impacts," Microsoft said in its 10-Q filing in January.
"These risks may arise from current copyright infringement and other claims related to AI training and output, new and proposed legislation and regulations, such as the European Union's AI Act and the US’s AI Executive Order, and new applications of data protection, privacy, intellectual property, and other laws."
But warning shots from the EU, US, and UK agencies may just lead to an AI jobs bonanza – for lobbyists. Last year, non-profit watchdog OpenSecrets reported that the number of lobbying entities focused on AI went from single digits in 2013, to 30 in 2017, and there were 158 of them in 2023. ®