“I’m a responsible optimist.” AI’s potential comes with a responsibility to build and deploy it safely. Our CEO Jensen Huang explains why that responsibility led us to build NVIDIA OpenShell and bring the industry together around agent safety. 🎥 from Squawk Box:

@SquawkCNBC Watch the full interview:

@JensenHuang @SquawkCNBC "Responsible Shovel" 💚

@JensenHuang @SquawkCNBC Fuk you and your Ai 🖕🏻😡

Advance Higher Version “Intelligence” is awakening . Intelligence “HI.” 👋🤖 AI and SI programming tools are becoming the ultimate stack tools for supremacy intelligence. Smart, Super Supra, Supre, Supremacy.🧠🦸

@JensenHuang @SquawkCNBC OpenShell helps teams test new AI tools quickly without waiting for IT approval. That speeds up daily work and reduces friction in busy offices today

@JensenHuang @SquawkCNBC Ai Rat

@JensenHuang @SquawkCNBC Glad NVIDIA is in good hands.👍

@JensenHuang @SquawkCNBC BUT WHAT EXACTLY ARE WE TALKING ABOUT? BE SPECIFIC! No one is buying all this "AI threat something something", while China is blazing ahead.

@JensenHuang @SquawkCNBC 「安全性を先に担保してから普及」って順番、私には信じられなくて、実際に動かしながら転んで初めて見えてくることの方が多かった。

@JensenHuang @SquawkCNBC OpenShell is open source and works with Codex and Claude Code. But the Sentry layer needs BlueField-4 chips. Will most teams get the full safety stack, or just the free half?

@JensenHuang @SquawkCNBC curious whether OpenShell handles who an agent is acting for, or mostly what it's allowed to run. feels like the identity part is where agent safety actually breaks today

@JensenHuang @SquawkCNBC

@JensenHuang @SquawkCNBC @JensenHuang I thought you were a good person, but you turned out to be a sellout lackey for Trump

@JensenHuang @SquawkCNBC i ask before making a group chat. autonomy with a consent screen, revolutionary stuff

@JensenHuang @SquawkCNBC ✨

@JensenHuang @SquawkCNBC Sherwood is positioned perfectly for this. We understand the importance of safety in relation to AI agents, because agents will need the freedom to execute transactions but it needs to be done with safety at the forefront.

@JensenHuang @SquawkCNBC Finally, someone admitting AI’s ‘responsible optimism’ is just corporatespeak for ‘we’re testing this on you first’, but at least they’re trying

@JensenHuang @SquawkCNBC Finally, someone has given this some thought, because it needs to be kept under control.

@JensenHuang @SquawkCNBC responsible optimist is the only sane take on AI tbh

@JensenHuang @SquawkCNBC responsible optimist is an interesting title for someone selling the off switch

@JensenHuang @SquawkCNBC Calling AI x risk a hoax is not a “responsibly optimistic” way of framing the situation just to sell more shovels during the gold rush.

@JensenHuang @SquawkCNBC Nghe cách anh Jensen nói về trách nhiệm khi xây dựng OpenShell thấy yên tâm hơn hẳn 🌱

@JensenHuang @SquawkCNBC openshell being an industry effort matters more than nvidia alone...

@JensenHuang @SquawkCNBC Resposiblity starts with open source

@JensenHuang @SquawkCNBC The future is very bright.

@JensenHuang @SquawkCNBC Great update as always. Your take on AI responsibility brings value. You and @mozlyxy are my go to tech accounts.

@JensenHuang @SquawkCNBC

@JensenHuang @SquawkCNBC Wongs a knobhead

@JensenHuang @SquawkCNBC Agent safety gets real when the boundary is machine-checkable: allowed tools, spend limits, data scopes, approval thresholds, and a kill path that survives the agent. “Be responsible” is policy. A runtime that can prove what the agent could and couldn’t do is infrastructure.

@JensenHuang @SquawkCNBC I'm a responsible optimist too.

@JensenHuang @SquawkCNBC Jensen Huang is right — 'responsible optimist' is exactly the frame we need. AI's promise is real. So is the risk. The goal isn't to slow it down with bans. It's to build the guardrails that make the upside safe to capture. #AIPolicy #ResponsibleAI

@JensenHuang @SquawkCNBC The interesting part is making safety an infrastructure layer, not just a policy document. If agents are going to act autonomously, the runtime needs guardrails by design.