US antitrust authorities have opened an investigation into the relationships between major artificial intelligence startups, such as OpenAI and Anthropic, creators of ChatGPT, and the tech giants who have invested billions of dollars in it.
The action targets Amazon, Google and Microsoft and their influence on the generative AI boom that has fueled demand for chatbots such as ChatGPT and other AI tools that can produce novel images and sounds.
“We are examining whether these links allow dominant companies to exert undue influence or gain privileged access in a way that could harm fair competition,” said Lina Khan, chair of the Federal Trade Commission. United States, in his opening speech at an AI forum on Thursday.
Khan said the market investigation would examine “investments and partnerships formed between AI developers and leading cloud service providers.”
The FTC said Thursday it had issued “mandatory orders” to five companies — cloud providers Amazon, Google and Microsoft, as well as AI startups Anthropic and OpenAI — requiring them to provide information on investments and partnerships.
The years-long relationship between Microsoft and OpenAI is the best known of the partnerships. Google and Amazon recently struck multibillion-dollar deals with Anthropic, another San Francisco-based AI startup formed by former OpenAI executives.
Google and Microsoft did not immediately respond to requests for comment. Amazon, Anthropic and OpenAI declined to comment.
The European Union and the United Kingdom have already indicated that they may also closely examine relations with Microsoft and OpenAI. The EU’s executive branch said in January it was checking whether the partnership could trigger an investigation under mergers and acquisitions regulations that would harm competition in the 27-nation bloc. Britain’s antitrust watchdog opened a similar investigation in December.
Antitrust advocates have welcomed the actions of the FTC and Europe in deals that some have called quasi-mergers.
“Large tech companies know they can’t buy the biggest AI companies, so they are finding ways to exert influence without formally calling it an acquisition,” Matt Stoller, director of research at the American Economic Liberties Project. “Law enforcement must intervene, and they do. »
Microsoft has never publicly disclosed the total amount of its investment in OpenAI, which CEO Satya Nadella described as a “complicated thing.”
“We have a significant investment,” he said during a November podcast hosted by technology journalist Kara Swisher. “It comes not just in the form of dollars, but also in the form of calculations and so on.”
OpenAI’s governance and relationship with Microsoft came into question last year after the startup’s board suddenly fired CEO Sam Altman, who was then quickly reinstated, in a turmoil that left the front pages of world newspapers. A weekend of behind-the-scenes maneuvering and a threat of a mass employee exodus championed by Nadella and other Microsoft executives helped stabilize the startup and led to the resignations of most of its previous board of directors .
The new arrangement gives Microsoft a non-voting board seat, although “we have absolutely no control,” Nadella said in Davos. Part of the complications that led to Altman’s temporary ouster involves the startup’s unusual governance structure. OpenAI began as a nonprofit research institute dedicated to the safe development of futuristic forms of AI. It is still governed as a nonprofit, although most of its staff work for the for-profit arm it created several years later.
Microsoft made its first billion-dollar investment in San Francisco-based OpenAI in 2019, more than two years before the startup introduced ChatGPT and sparked global fascination with advances in AI.
Under the deal, the Redmond, Wash., software giant would provide the computing power — such as from one of its data centers in rural Iowa — needed to drive AI models on huge amounts of human-written text and other media. In turn, Microsoft would gain exclusive rights to much of what OpenAI has built, allowing the technology to be integrated into a variety of Microsoft products.
Nadella in January compared it to a number of Microsoft’s long-standing business partnerships, such as with chipmaker Intel. Microsoft and OpenAI “are two different companies, accountable to two different stakeholder groups with different interests,” he told a Bloomberg reporter at the World Economic Forum in Davos, Switzerland.
“So we build the calculation. They then use computation to perform the training. We then take that and put it into products. And so, in some sense, it’s a partnership based on each of us really strengthening what each other is doing and ultimately being competitive in the marketplace.
The FTC has been signaling for nearly a year that it is working to track and stop illegal behavior in the use and development of AI tools. Khan said in April that the US government would “not hesitate to crack down” on harmful business practices involving AI. One of the most concerning targets is the use of AI-generated voices and images to power fraud and phone scams.
But increasingly, Khan has also made clear that it is not just harmful applications, but also the broader consolidation of market power among a handful of AI leaders that deserves government scrutiny. “Companies could take advantage of this critical market moment to exploit anticompetitive tactics to consolidate their dominant position and block competition,” the FTC said in a preview of Thursday’s forum.
Companies have 45 days to provide the FTC with information including their partnership agreements and the strategic rationale behind them. They are also asked for detailed information on decision-making around product launches and the key resources and services needed to create AI systems.
Kelvin Chan, AP economics editor in London, contributed to this report.