Salesforce’s Dreamforce conference opened Tuesday with Gwen Stefani performing “Underneath It All” before CEO Marc Benioff took the stage. Benioff projected a graph showing Salesforce’s projected $46 billion in annual revenue by 2027, citing demand for AI-powered products. The conference also hosted a debate over whether the AI industry should slow down to avoid catastrophic risks.
Coxon’s Warning and the Debate
AI researcher Jacob Coxon resigned from Anthropic last week, publicly warning that companies racing toward self-improving AI systems were “gambling with our lives.” His warning framed the stakes for the panelists who followed.
Anthropic CEO Dario Amodei urged world leaders to help “pace the frontier” of AI development. OpenAI CEO Sam Altman endorsed the idea. Amodei’s proposal to pace the frontier drew support from Altman, though the source does not say Altman agreed with Amodei’s reasoning.
David Sacks, a White House AI adviser, disagreed. He dismissed Amodei’s warnings as “just another bid for regulatory capture.” The split between the two camps was sharp: one side wants governments to slow the pace of AI, and the other wants them to stay out entirely.
Trump’s Hoax Claim and the Push for Regulation
President Trump called fears about existential AI risk “a hoax.” That position contrasted with the panelists who spoke at Dreamforce.
Congress is pushing legislation to regulate the AI industry despite the political climate. The push for regulation continues even as some politicians deny the risks exist.
The contrast between the conference and Washington was stark. While executives argued over safety standards, lawmakers were drafting rules that could reshape the industry.
Huang’s Engineering Answer
Nvidia CEO Jensen Huang appeared onstage, arguing AI safety is an engineering problem that companies can police themselves. Huang said: “We don’t need any new laws, we don’t need any new regulation.”
Huang’s position was the clearest rejection of Amodei’s proposal. He argued that companies can build safer systems without government intervention.
Kapoor’s Liability Case
Francisco “Sayash” Kapoor, an incoming UC Berkeley computer science professor and coauthor of AI Snake Oil, made a separate argument. Kapoor argues AI labs should be held liable for the harms their technology causes.
Kapoor noted OpenAI faced minimal consequences after the Hugging Face hack. He said: “The consequences for OpenAI for the Hugging Face hack were basically close to zero.”
Kapoor’s point was that the current system fails to hold labs accountable. If companies know they will face few consequences, they have little incentive to fix problems.
The Hugging Face Incident
Amodei made a reference to OpenAI’s Hugging Face incident during a discussion with Benioff. The hack was a central example in the debate.
The incident raised questions about how companies respond when things go wrong. It also drew attention to the gap between what labs claim and what they deliver.
Kapoor’s research, including work with Princeton professor Arvind Narayanan, has questioned the claims of AI labs.
The Debate at Dreamforce
The conference felt like a megachurch with a packed audience. The debate over AI safety brought together executives pitching enterprise tools with academics questioning the industry’s direction.
Kapoor and Narayanan are coauthors of the book AI Snake Oil and write a Substack called AI as Normal Technology. They published one of their longest essays yet, which tries to make sense of “loss-of-control incidents,” referring to events such as OpenAI accidentally letting its agents hack into Hugging Face.
Key facts box:
– Salesforce projects $46 billion in annual revenue by 2027
– Jacob Coxon resigned from Anthropic last week
– Amodei urged world leaders to “pace the frontier”
– Huang said: “We don’t need any new laws, we don’t need any new regulation”
– Kapoor said the consequences for OpenAI were “basically close to zero”
Order of events:
1. Stefani opened the conference with her performance
2. Benioff presented Salesforce’s revenue projection
3. Coxon’s resignation was announced
4. Amodei urged pacing the frontier
5. Altman endorsed Amodei’s position
6. Sacks dismissed Amodei’s warnings
7. Huang argued AI safety is an engineering problem
8. Kapoor argued for liability
The debate at Dreamforce showed the AI industry is deeply divided over how to manage risk. Companies that want to move fast will push back against regulation. Companies that worry about risks will argue for caution.
The conference turned into a battleground for the future of AI.
Where the paper stands
The paper backs Salesforce’s push for AI growth and is against Anthropic CEO Amodei’s call for government intervention. The technology itself is not the danger; the danger is letting the biggest firms decide who gets to use it.
Salesforce’s $46 billion revenue projection shows real demand for AI-powered products. Benioff’s presentation at Dreamforce was a vote of confidence in the industry’s future, not a bet against safety. The paper supports narrow rules against direct harm, such as forcing companies to disclose safety failures they hid. Salesforce’s push is about meeting that demand without freezing the market.
Amodei’s proposal to “pace the frontier” asks governments to step in and slow the pace of AI development. That is the opposite of what the paper supports. Broad new rulebooks, agencies widening their reach, and rules written with the biggest firms’ help all protect incumbents and raise the cost of entry for everyone else. Amodei’s proposal fits that pattern. The paper opposes it.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.