Nvidia, Palantir said to restrict AI model use

by Alimat Aliyeva

Nvidia Corporation, Palantir Technologies, and Booz Allen Hamilton have started limiting or completely eliminating the use of the artificial intelligence models from Anthropic and OpenAI amid growing fears of those models learning from their customers' intellectual property, the Information reported on Monday.

According to the report, while OpenAI and Anthropic claim that their models are not trained using the input from companies with enterprise contracts, they use "some metadata" from those customers, with some users reportedly saying that the AI firms' disclaimers about the data retention policies are "not clear enough."

The report also stated that Microsoft, Inc. has been using the allegations made against Palantir and OpenAI to bring the corporate customers to its own AI models by offering its own "isolated" cloud environments and even a system that runs AI models on privately managed server which do not transfer user data to external AI providers.