Dreamforce 2026 and the dangers of AI

AI safety has dominated the public discourse over the last month, with executives and founders of some of the most popular AI companies sounding the alarm themselves. And the topic was a key talking point at Salesforce's annual event, Dreamforce.

Salesforce CEO Mark Benioff was joined on stage by Nvidia's Jensen Huang and Anthropic's Dario Amodei, with both discussing the current state of AI and also Salesforce's own AI offerings.

ITPro's news and analysis editor Ross Kelly joined the pod live from San Francisco to talk AI safety, Claudforce, Koa, and all things Salesforce Dreamforce.

Highlights

"We need to remember that Salesforce and Anthropic have just announced Claudeforce as well, which is essentially deeper integration of Claude across Salesforce's product ecosystem, but also you know new Salesforce plugins for Claude as well. I think the fact that AI safety came up, everybody really sprung up in the crowd. I think if you look at a lot of the headlines, that's definitely dominated the agenda from a lot of the press, ourselves included. Obviously, it wasn't just Amodei that discussed; Jensen Huang obviously touched on it, and then after the keynote, during the press scrum with Mark Benioff."

"The responsibility lies firmly with these model developers slowing down the entire industry because these companies are concerned about, you know, the risks posed by their models; that's not an option, really. He specifically highlighted testing. He said testing environments need to be built in a good way, basically to avoid incidents like the Hugging Face attack.

Links

• Dreamforce 2026 showed AI safety is the new big tech battleground
• Marc Benioff and Jensen Huang aren't sold on the 'SaaSpocalypse'
• Salesforce teams up with Nvidia to launch Koa, a dedicated CRM reasoning model