Salesforce CEO Marc Benioff's message to companies interested in generative AI is clear: The technology can't be trusted yet. Unless his company is selling it.
AI was the resounding theme at the company's annual flagship conference, Dreamforce, which I attended in San Francisco this week. Salesforce, which makes cloud-based software for customer relationship management, billed this year's Dreamforce as "the world's largest AI event" and began the week with the announcement of its new generative AI product, Einstein 1.
Benioff spent much of the conference issuing ominous warnings about the perils of generative AI while simultaneously touting his own company's new technology as ethical and secure.
Onstage at the Yerba Buena Center for the Arts, he prodded celebrities, politicians, and business leaders — including OpenAI CEO Sam Altman — to share their concerns about the existential risks of feeding intelligent machines vast amounts of personal data before reminding the audience that Salesforce was setting the standard for handling data ethically and responsibly.
After two days of listening to this, I started to wonder how Benioff could be so sure his large language models aren't behaving as badly as everyone else's and training on copyright-protected data.
"We're not scraping the internet with our models, if that's your question," Benioff told me Wednesday afternoon.
Salesforce has "an open philosophy" regarding the development of large language models, or LLMs, Benioff said, building some on top of preexisting models and building others from scratch. For example, CodeGen, an LLM released by Salesforce in 2022, was trained from scratch using Apex, an internal programming language.
"It's inside our core system," Benioff said. "It's not being generated by the computer."
Benioff, who bought Time magazine in 2018, has a particular distaste for copyright infringement.
"As a media owner, it's a major issue, because I do go to the models, and I'll find material from Time magazine in there and go, 'Wait a minute, that's my content,'" Benioff said.
"Media companies are starting to wake up and realize a lot of their information has been stolen — probably some of yours, too," he added, giving me a pointed look.
Before I could determine whether the look was made in jest, Kathy Baxter, Salesforce's principal architect of responsible AI and tech, interjected.
Baxter said that Salesforce only uses customer data with consent and performs legal reviews before using any open-source data for AI model training.
"I'm very proud of all the models that we are creating," Baxter said. "We can be extremely transparent in that data, and you can stand by it and know that it is not copyrighted. Never."
Got a tip for this reporter? Contact Ellen Thomas at ethomas@insider.com or reach out on Signal at (646) 847-9416. Use a non-work device.
