Musk's Quip Highlights Benioff's AI Tool Switch From ChatGPT to Grok

Marc Benioff stood on stage at Dreamforce 2025, Salesforce's flagship conference, and told the crowd something that would ripple across Silicon Valley within hours: ChatGPT had refused to do what he asked. He wanted to edit a photograph—specifically, to remove branded necklaces from an image of himself and Sam Altman, OpenAI's CEO. ChatGPT declined, citing copyright concerns tied to the branded elements in the shot. So Benioff turned to Grok, Elon Musk's AI tool, which completed the task without hesitation.

The moment might have stayed confined to the conference hall, a minor anecdote about AI tool limitations, except that a popular X account called DogeDesigner shared video of Benioff's remarks. Musk saw it and replied with two words: "Haha cool." That casual response, posted by one of the world's most followed accounts, accumulated more than 90,000 likes and sparked over 3,000 replies. What might have been a throwaway comment became a small but visible crack in the facade of enterprise AI loyalty.

The timing made the moment sharper. Just months earlier, in October 2025, Benioff had announced an expanded partnership between Salesforce and OpenAI. The deal integrated Salesforce's Agentforce 360 platform with ChatGPT, allowing enterprise customers to access OpenAI's GPT-5 models directly within Salesforce products. At that announcement, Benioff had spoken warmly about the collaboration, noting that consumers already benefited from ChatGPT's instant recommendations and insights, and now enterprises could deliver that same capability. He had been, by any measure, ChatGPT's most vocal corporate champion.

But by November 2025, something shifted. Benioff tried Google's newly released Gemini 3 model and posted his reaction on X with unmistakable enthusiasm: "Holy s—. I've used ChatGPT every day for 3 years. Just spent 2 hours on Gemini 3. I'm not going back." He elaborated on what he saw—reasoning, speed, image and video capabilities, all sharper and faster than what he'd been using. "It feels like the world just changed, again," he wrote. The language was the language of conversion, not incremental improvement.

What emerged from these moments was a portrait of the AI market in early 2026: volatile, competitive, and capable of shifting the allegiances of even its most prominent advocates. Benioff's public journey from ChatGPT evangelist to Gemini 3 convert to Grok user (however briefly) illustrated that no AI tool had yet secured the kind of lock-in that might have seemed inevitable a year earlier. Enterprise leaders were testing, comparing, and moving. Musk's two-word reply to the Grok anecdote was less about endorsement and more about acknowledgment—a nod to the fact that his tool had done the job when the market leader would not.

The Dreamforce stage also hosted another moment of recalibration from Benioff. He apologized for earlier statements in which he had suggested that President Trump should deploy the National Guard to San Francisco to address safety concerns, despite opposition from local officials. Benioff, whose company is San Francisco's largest private employer and who holds considerable sway in city politics, said he had reconsidered after listening to fellow San Franciscans and local leaders. He clarified that he did not believe the National Guard was necessary. The apology suggested a CEO navigating not just technological shifts but political ones, recalibrating his public positions as circumstances evolved. In the span of a single conference, Benioff had moved on multiple fronts—from AI tools to city governance—each shift a small signal that the ground beneath these decisions was less stable than it appeared.