What exactly does it mean to "slow down” artificial intelligence? How exactly would that work? ABC News’ Elizabeth Schulze breaks down why tech CEOs are urgently warning about the potential harm of AI and the growing questions over how to regulate it.

“Slow down AI” sounds simple, but the debate is really about putting more safety testing and oversight between increasingly powerful models and the public. The technology could keep advancing while the rules and safeguards catch up.

Altman is a danger.

The last time this crowd tried to run the truth, they built the Disinformation Governance Board. Now some of the same political team wants regulate us again with AI, pausing data centers, banning superintelligence, and standing up a new federal agency to approve models before they ship. That's not safety, it's the same play with a fancier name.

The whole "urgent warnings" angle feels like every major industry leader is saying the same thing right now to look proactive. What specific guardrails do they think actually work outside of just issuing PR statements?

The better question isn't "slow down" — it's where and how? Slow down deploying AI in health records without audits? Yes. Slow down the research before we understand alignment? Reasoned debate. Slow down everything indefinitely? Economically and strategically impossible. The real answer: guardrails, not bans. #AIPolicy #TechPolicy

Vitalik offers a different perspective on AI safety: