> ## Content Index
> Fetch the complete content index at: https://wire.fourthweb.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# Nvidia CEO Tells OpenAI's Altman AI Isn't Slowing Down
- URL: https://wire.fourthweb.ai/nvidia-ceo-tells-openais-altman-ai-isnt-slowing-down/
- Published: 2026-09-17T21:31:03.000Z
- Updated: 2026-09-17T21:31:06.000Z
- Description: The Dreamforce stage lit up with the kind of argument VCs usually have in private: slow down or scale up, and who gets to decide.
- Author: Travis Wright
- Tags: AI Agent Economy, Agentic Workflows, AI Agents, AI Infrastructure, Compute Wars, AI Governance, DeFi, OpenAI, Anthropic

**The Dreamforce stage lit up with the kind of argument VCs usually have in private: slow down or scale up, and who gets to decide.**

### The Summary

- [OpenAI, Anthropic, and Nvidia CEOs turned Salesforce's Dreamforce into an impromptu debate on AI development velocity](https://www.wired.com/story/are-rogue-ai-agents-really-just-a-cybersecurity-problem/?ref=wire.fourthweb.ai), with the conversation quickly spiraling into territory none of them expected.
- [By calling their coordination a "slowdown" instead of security standards development, AI labs may have triggered antitrust scrutiny](https://www.wired.com/story/the-ai-slowdown-is-an-antitrust-mess/?ref=wire.fourthweb.ai) that could dog them for years.
- The framing matters more than the intention. What sounds like responsible restraint to AI safety advocates looks like cartel behavior to regulators.

### The Signal

Salesforce's annual developer lovefest became the unlikely venue for AI's C-suite to air a disagreement that's been simmering in boardrooms for months. The core question: should the industry tap the brakes on capability advances to get security and safety measures in place, or is "slowing down" just code for incumbent protection?

[The debate itself isn't new, but having it unfold publicly at Dreamforce](https://www.wired.com/story/are-rogue-ai-agents-really-just-a-cybersecurity-problem/?ref=wire.fourthweb.ai) signals something shifted. These aren't scrappy startups anymore. [OpenAI](https://wire.fourthweb.ai/tag/openai/), [Anthropic](https://wire.fourthweb.ai/tag/anthropic/), and [Nvidia](https://wire.fourthweb.ai/tag/nvidia/) together represent the controlling interests in the agent economy's infrastructure layer. When they talk about coordination, antitrust lawyers start billing hours.

> "By framing their efforts as a 'slowdown' rather than an industry-wide push for better security standards, AI labs may have set themselves up for years of regulatory headaches."

Here's the regulatory trap [the labs walked into](https://www.wired.com/story/the-ai-slowdown-is-an-antitrust-mess/?ref=wire.fourthweb.ai): if you're coordinating on development timelines, you're potentially fixing market competition. It doesn't matter if your stated intent is safety. The FTC and DOJ don't grade on good intentions. They look at market effects. Three of the most powerful companies in AI discussing when to ship what features? That's the definition of antitrust exposure.

The smarter play would have been to frame it as security standards development from day one. Industries coordinate on security baselines all the time:

- Payment processors built PCI-DSS together without antitrust issues
- Cloud providers coordinate on SOC 2 compliance frameworks
- Even social media platforms share threat intelligence through structured programs

But "AI slowdown" sounds like an agreement to restrict output. Which is exactly what competition law prohibits.

The timing makes it worse. We're entering the phase where [autonomous agents will start handling real transactions](https://www.wired.com/story/are-rogue-ai-agents-really-just-a-cybersecurity-problem/?ref=wire.fourthweb.ai), moving money, signing contracts, making purchases. The cybersecurity implications are real. But if the companies building these agents have already been publicly discussing coordinated slowdowns, any technical standard they propose now will be viewed through an antitrust lens.

### The Implication

If you're building on top of foundation models, this semantic fumble matters for your roadmap. Expect capability releases to become less predictable as labs navigate the regulatory scrutiny they created for themselves. The safest bet for labs will be to ship security features quietly and independently, which means less coordination on the standards that would actually make agent-to-agent interaction safer.

For enterprises deploying agents, the takeaway is simple: don't wait for industry standards that may never materialize cleanly. Build your own security frameworks now. The labs might have the capability to set good standards, but they've lost the regulatory freedom to coordinate on them openly.

### Sources

[Wired AI - The AI 'Slowdown' Is an Antitrust Mess](https://www.wired.com/story/the-ai-slowdown-is-an-antitrust-mess/?ref=wire.fourthweb.ai) | [Wired AI - The AI Slowdown Debate Crashed Salesforce's Party](https://www.wired.com/story/are-rogue-ai-agents-really-just-a-cybersecurity-problem/?ref=wire.fourthweb.ai)