This deal would primarily involve the use of OpenAI models from Microsoft’s Azure cloud, The Information reported Wednesday (Nov. 18), citing two sources with knowledge of the situation.
The report notes that the move to expand ByteDance’s already-significant Microsoft contract has upset OpenAI.
The artificial intelligence company has complained to Microsoft in recent months that it believes Chinese companies are “distilling” its AI, or running OpenAI models to develop their own competing AI models, according to The Information’s sources.
OpenAI itself blocks most Chinese firms from using its AI, but Microsoft — which has the rights to sell OpenAI’s technology — doesn’t have the same prohibitions, the report said.
At least some of the deal with ByteDance represents business previously provided through Anthropic, one of the sources said. Anthropic doesn’t permit its products to be sold through cloud platforms such as Azure to customers in prohibited countries, and it has increasingly blocked use of its technology via other channels, The Information said.
Anthropic had announced last September that it was cutting off Chinese customers that accessed its services using subsidiaries incorporated in other nations. A July Financial Times report said the Claude maker was increasing enforcement to shut down Chinese use through these loopholes that permitted Chinese customers to still use its AI tools, including through foreign subsidiaries using Azure.
Microsoft is also pursuing deals to sell OpenAI models to other Chinese companies that lost access to Anthropic, including Alibaba, a source familiar with the situation told The Information.
In other Microsoft news, PYMNTS wrote last week about the fallout from the widespread outage of the company’s Outlook system.
“The first-order implication of the outage is that many of the cloud-based systems companies rely on for efficiency can also introduce operational dependencies that are expensive to manage,” that report said.
The April PYMNTS Intelligence report “Scale Amplification: How Revenue Amplifies Agent-Driven Identity” showed that large enterprises, with their bigger digital footprints, can be more at risk of new AI-powered attacks favored by fraudsters.
“Historically, companies could evaluate software investments primarily through licensing costs, implementation expenses and anticipated productivity improvements,” PYMNTS added. “Today, the financial equation includes integration maintenance, security testing, change management and the potential cost of interrupted operations.”