I’ve been tracking this AI safety chorus carefully this week, starting Sep. 14. It started with Anthropic CEO Dario Amodei’s essay calling for a moderated pace of frontier model development.
Then Broadcom CEO Hock Tan said it changed nothing about his revenue targets. Then CrowdStrike CEO George Kurtz argued that slowing what comes next doesn’t secure what’s already out there.
Now Marc Benioff, Salesforce (CRM) co-founder and CEO, has joined the conversation with a different warning. He didn’t call for a slowdown or dismiss the concerns either.
Instead, sitting down with Jim Cramer at Dreamforce on Sep.15, he pointed to an industry that moved fast, shaped society, and left a trail of damage that regulators are still sorting through years later.
A lot of companies got hurt, a lot of individuals got hurt through social media. We don’t want that to happen in AI.
Coming from the man who called Facebook “the new cigarettes” back in 2018, that framing carries weight, according to TheStreet‘s reporting.
Also Read: Salesforce latest news and stories
Why Salesforce’s Benioff warning lands differently than the others
Amodei wrote from the perspective of someone building the most powerful models in existence and genuinely worried about what will happen next. Tan responded as a hardware supplier whose entire business depends on continuing compute demand.
Kurtz spoke from the front lines of cybersecurity, where he says the threat has already evolved faster than most organizations can respond.
Benioff is speaking from a different vantage point entirely. He runs the world’s largest customer relationship management (CRM) platform. Its software sits amidst thousands of enterprises managing billions of customer relationships.
If AI goes badly for end users, Salesforce feels it. If AI destroys trust at the application layer, Salesforce’s customers bear the consequences.
More AI:
Nvidia just made a move Wall Street wasn’t ready for
Microsoft just took sides in AI policy fight
OpenAI just disclosed something genuinely alarming
“The heart of ethics is responsibility,” Benioff said on Mad Money. “We want AI to take care of these things and be held responsible, and take care of these actions and be ethical.”
He stopped short of endorsing a slowdown or criticizing specific companies. But the social media comparison is pointed. The argument he’s making is that industries that move fast without building accountability mechanisms tend to create damage that outpaces their ability to repair it. He wants AI to be different.
In late Aug. 2026, Salesforce unveiled Claudeforce — a plugin that allows clients to use Anthropic’s Claude to access Salesforce customer data and automate tasks like composing emails and updating records.
Roughly three weeks later, Amodei appeared on Benioff’s Dreamforce keynote stage. They are both advocating for caution, and both publicly endorsing each other’s products.
What the Salesforce numbers show about the AI tailwind Benioff is navigating
Salesforce reported Q2 fiscal 2027 results on Aug. 26 that Benioff called “one of our best quarters ever.”
Revenue hit $11.3 billion, up 11% year-over-year (YoY)
Current remaining performance obligations reached $33.5 billion, up 14% YoY
Free cash flow came in at $1.1 billion, up 81% YoY
The AI-specific metrics are where the growth is most visible. AI and data product ARR is approaching $4 billion. Net new annualized order value (NNAOV) growth is the strongest it’s been in four years, according to CFO Robin Washington‘s comments in the Q2 statement.
Salesforce raised full-year fiscal 2027 revenue guidance to $46.1-$46.4 billion, up 11-12% YoY, incorporating two pending acquisitions — Contentful and Fin — expected to close in Q3.
Related: Salesforce just got the signal bulls were waiting for
I find AIforce — Salesforce’s enterprise AI harness connecting agents, models, and interfaces to customer data — to be the most strategically significant product in the portfolio right now.
Benioff described it as “unlocking the data, workflows, business logic, actions, and governance inside Salesforce and making it available to every agent, model, and interface,” according to the company statement. That’s a Salesforce positioning play for the agentic AI era.
On the stock performance. CRM is down 4.94% year-to-date and up only 5.58% over the past year, according to Yahoo Finance. But it has recovered more than 50% from its 52-week low of $146 in late June. Overall, Enterprise software broadly surged Sep. 14 when Amodei’s essay hit.
In late Aug. 2026, Salesforce unveiled Claudeforce — a plugin that allows clients to use Anthropic’s Claude to access Salesforce customer data and automate tasks like composing emails and updating records. Tigarto Via Shutterstock
Why the chorus matters for enterprise software investors
My read on this week’s AI safety debate, having covered it from Amodei to Tan to Kurtz to Benioff, is that the narrative is shifting in a way that’s structurally positive for enterprise software companies.
For much of the first half of 2026, the dominant fear was that AI would hollow out enterprise software by automating the tasks that platforms like Salesforce charge subscription fees to enable. That fear drove CRM to its June lows.
What’s emerging now is a different picture. AI agents need governance. They need data. They need workflow infrastructure. They need the kind of enterprise-grade accountability frameworks that Benioff has been building for 20 years. The Claudeforce partnership isn’t Anthropic competing with Salesforce but Anthropic needing what Salesforce already has.
Benioff’s social media warning is a genuine ethical stance too. It’s also, perhaps coincidentally, a very good business argument for why enterprises will keep paying for platforms that add structure and accountability to AI deployments rather than simply replacing human workflows with uncontrolled agents.
The heart of ethics is responsibility. That’s also in enterprise software, the heart of the value proposition.
Related: Salesforce just gave Wall Street a reason to believe its AI story