Fears about artificial intelligence (AI) going out of control are not new.

But these fears are getting more attention now.

More people are warning about AI systems that do bad things.

Some top AI bosses recently agreed on one thing.

They said companies should slow down together.

They also said countries should work together.

This can protect people from the worst problems.

Other tech leaders disagree.

They say their companies can keep AI safe alone.

Dario Amodei is the CEO of Anthropic.

He gave the most detailed plan to slow down.

He said top labs should let outside experts come in.

These experts should have long-term access.

They should get offices, badges and laptops.

Anthropic and OpenAI both agreed to do this.

Amodei also asked for government rules.

He said top AI companies, democratic governments and authoritarian governments should all work together.

He said this is hard.

He talked about China especially.

But he said, "we owe it to humanity to try."

Sam Altman is the CEO of OpenAI.

He supports slowing down and working together more.

OpenAI has asked for required national safety rules.

It has also supported state laws.

Chris Lehane is OpenAI's chief global affairs officer.

He said the company would support international ways to measure AI power and manage risk.

He said this is true even if it slows down AI models.

Altman said slowing down does not mean stopping.

He said, "no amount of American competitive pressure should justify recklessness."

Mark Zuckerberg is the CEO of Meta.

He disagreed.

He said each company should make its own technology safe.

Jensen Huang is the CEO of Nvidia.

He also criticized calls to slow down.

He said the market is enough.

He said no new laws are needed.

Elon Musk supported Amodei.

He wrote, "Dario is right."

He said competitors should test each other's models.

Demis Hassabis is from Google DeepMind.

He agreed with Amodei's direction.

He proposed a group paid by the industry to set AI standards.

Satya Nadella is from Microsoft.

He said AI must help people and stay under human control.

But he warned that a few groups cannot control everything.

Mustafa Suleyman is the CEO of Microsoft AI.

He said he worries that treating models like people makes safety risks worse.