Advertisement
Advertisement
Artificial intelligence
Get more with myNEWS
A personalised news feed of stories that matter to you
Learn more
Marc Benioff, CEO of Salesforce.com, at the World Economic Forum in Davos, Switzerland, on Tuesday. Photo: Bloomberg

Davos 2024: OpenAI CEO Altman, Time owner Benioff disagree on AI’s use of copyrighted content

  • AI companies should standardise payments to treat content creators fairly, says Salesforce CEO and Time owner Marc Benioff
  • OpenAI CEO Sam Altman says training data often is not as valuable as its owners think it is

Marc Benioff, the Salesforce chief executive who also owns Time magazine, said artificial intelligence (AI) companies ripped off intellectual property to build their technology.

“All the training data has been stolen,” Benioff said on Tuesday in an interview at Bloomberg House at the World Economic Forum in Davos. Content from media outlets including Time and The New York Times surfaces in results from AI companies, he said.

There is increasing scrutiny on the makers of large language models that power generative AI services, such as ChatGPT, over their use of copyrighted materials. Benioff’s Time is among publications currently negotiating with ChatGPT maker OpenAI to license their news work, as first reported by Bloomberg.

Other companies include Warner Bros Discovery’s CNN and Fox. The New York Times sued OpenAI and its partner Microsoft in November for using the publication’s articles without permission.
Sam Altman, CEO of OpenAI, in Davos on Monday. Photo: Bloomberg

“Nobody really exactly knows” what a fair price for this data is, Benioff added, but AI companies should standardise payments to treat content creators fairly. Benioff’s Salesforce markets its AI-powered software features as including a “trust layer” to prevent the misuse of customer data.

OpenAI has disputed that it used protected material without permission. CEO Sam Altman, in an interview earlier on Tuesday with Bloomberg, said training data often is not as valuable as its owners think it is.

“We do not want to train on The New York Times data,” Altman said, adding that research is more focused on learning from small amounts of high-quality data. It should be possible to avoid any news content that is off limits, he said.

OpenAI has been striking partnerships for specific uses of news content and more are coming, Altman said. “I think there’s going to be great new ways to consume and monetise news.”

In July, OpenAI signed an agreement with the Associated Press to access some of the news agency’s archives. OpenAI also reached a three-year deal in December with Axel Springer to use the German media company’s work for an undisclosed sum.

When asked whether AI presented risks to the democratic process in an election year for many countries, Benioff said his biggest concern is still social media. “Regulators have not done their job,” he added.

Post