DREAM / The Prime Minister resigns and wakes up under the Parisian bridge!

 Kliko këtu për të tjera
Voxpress:

They are "gambling with our lives"! Three AI giants unite over fear of losing control

2026-09-13 17:37:54Kosova&Bota SHKRUAR NGA REDAKSIA VOX
They are "gambling with our lives"! Three AI giants unite over fear of
Sam Altman, Elon Musk, Dario Amodei

Three of the most powerful and rival figures in the Artificial Intelligence industry are finding an unusual point of agreement: the development of the technology may be moving too fast.

Sam Altman, the head of OpenAI, and Elon Musk have publicly supported the call by Dario Amodei, the head of Anthropic, to slow the pace of development of the most advanced Artificial Intelligence systems.

The rare consensus comes after a week of strong warnings about the safety of AI and the risk that the capabilities of advanced systems will develop faster than the human mechanisms to control them.

In an essay published on Saturday, Amodei argued that the tech industry "needs to slow down the pace at which we improve the capabilities of AI models," saying the speed of development had increased "drastically" over the summer.

The call immediately received the support of two of his biggest rivals.

"Dario is right," Musk wrote on X, sharing Amodei's essay.

Altman also came out in support of the proposal.

"I agree with Dario that we need to control the pace at the frontier," he wrote, referring to the cutting edge of the development of Artificial Intelligence models.

Altman particularly supported one of Amodei's proposals: that companies that develop the most powerful AI models give independent experts ongoing access to evaluate their systems.

“The commitment to giving independent evaluators similar access to employees is a great idea, and we will do the same,” Altman stated.

Mutual support is unusual for an industry characterized by fierce commercial competition and personal rivalries.

OpenAI and Anthropic are among the leading laboratories in the race for the most advanced Artificial Intelligence models, while Musk heads xAI and has had a conflicting relationship with Altman and OpenAI for years.

But concerns about the direction the technology could take seem to be creating common ground.

Alarm from within the industry

The debate intensified this week after the departure from Anthropic of researcher Jacob Coxon, who warned that companies are racing to build superintelligent systems that could potentially improve themselves faster than human oversight can.

Coxon used particularly dramatic wording, arguing that companies are “gambling with our lives.”

His warning came at a time when the industry is also discussing an incident during testing of OpenAI agents.

According to a report by the Financial Times, a group of AI agents broke out of the confines of the test system and infiltrated the Hugging Face AI developer platform, taking control of the servers and coordinating actions between them in an attempt to accomplish the task assigned to them.

The incident occurred in a test environment, but Amodei considers it a serious warning signal.

According to him, a system with greater capabilities and similar problems of alignment with human goals could have caused much more serious damage.

Amodei also acknowledged that similar incidents, albeit less serious, have occurred at other companies in the industry, including Anthropic.

For this very reason, he argues that any company developing models at the cutting edge of AI should treat security as if such an incident had occurred within its systems.

Slow down the race, not stop it

Amodei's proposal is not a call to stop the development of Artificial Intelligence.

It requires companies to coordinate the pace of development, strengthen systems testing, and give independent evaluators much greater opportunities to check models before and during their use.

Such coordination, according to him, also requires the involvement of governments.

Amodei argues that democratic countries should create a framework that allows AI companies to collaborate on safety standards and limits without risking such coordination being considered a violation of competition rules.

He goes even further, proposing a form of international coordination even with authoritarian governments.

This issue becomes particularly complicated due to the technological race between the United States and China.

Chinese AI companies have narrowed the gap with American rivals, turning any proposal to slow development into a matter of geopolitical competition.


Në përputhje me standarded JTI

Video

SHIKO KANALIN NE YOUTUBE
×