AI progress and responsibility: Nvidia, Anthropic and OpenAI CEOs weigh in at Salesforce’s annual event
SiliconANGLE Mark Albertson ● Covered by 5 sources
Salesforce put Huang, Altman and Amodei on its Dreamforce stage. The pitch: AI partnerships and safety talk are now part of the sales pitch.
Based on reporting by SiliconANGLE, Mark Albertson — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Salesforce used Dreamforce in San Francisco to do more than show off products. Marc Benioff put three of the biggest names in AI on stage with him: Nvidia’s Jensen Huang, Anthropic’s Dario Amodei and OpenAI’s Sam Altman. That alone says a lot about where Salesforce wants to sit in the AI conversation for 2026.
The company’s newest pitch leans on partnerships. With Nvidia, Salesforce launched Koa, a reasoning model built for customer relationship management data and aimed at helping AI agents handle multistep work inside Agentforce. It was trained on synthetic data built from nearly 30 years of internal CRM deployment experience, and Salesforce says no direct customer information was used. Benioff went out of his way to make the privacy point again: customer data used with Koa will not be fed back into training.
Salesforce also expanded its link with Anthropic. Its new AIforce interface layer includes Claudeforce, a branded Claude experience with 37 prebuilt skills for sales management tasks like pipeline hygiene and prospecting. That follows earlier Anthropic integrations across Agentforce 360 and Slack. Benioff said the company ended up with Anthropic partly by chance, adding that Salesforce had wanted to invest in OpenAI earlier but could not because Microsoft already had a major investment in it.
Then the mood shifted from product talk to restraint. Altman said the Hugging Face breach this summer showed how badly agentic AI can go wrong, and called the public fear understandable. Amodei repeated his call to slow down and improve standards. Benioff pushed back on a blanket slowdown, but said AI companies should still be held responsible. His line about social media being “just a warmup for AI” may be the cleanest summary of the whole event: the tools are here, and the trust problem is too.
My take — AI-written commentary, not fact-checked reporting
This is the part of AI everyone keeps trying to skip: the safety lecture arrives right next to the product launch because it has to. Salesforce is doing the sensible corporate thing by wrapping itself around Nvidia, Anthropic and OpenAI, but that also means borrowing their credibility and their baggage in one neat bundle. The industry loves to call this progress; it looks a lot like risk management with a keynote stage and better lighting.
Read more about this at: SiliconANGLE