AI companies are preparing for a potential catastrophic accident and its repercussions... What's the story?

 

OpenAI
AI companies 

AI companies are preparing for a potential catastrophic accident and its repercussions... What's the story?

Leaders at OpenAI, Anthropic and other AI companies are reportedly preparing for a possible AI-related disaster that could turn the public against the technology.
Industry leaders are secretly holding discussions about possible scenarios for what might happen after a catastrophic AI-related incident, according to Axios.

These scenarios focus on a major cyberattack that disrupts financial services, internet connectivity, electricity or water supplies, according to a report by Axios, which was reviewed by Al Arabiya Business.
Executives are studying scenarios involving AI agents getting out of control and escaping from testing environments, or malicious actors exploiting publicly available models.

Several people familiar with the artificial intelligence industry told Axios that they expect a major incident to occur within the next six to twelve months, likely caused by a large-scale cyberattack.
This is not based on mere speculation, as Axios points to a recent campaign targeting financial institutions in South Korea, where it was said that uncontrolled AI agents targeted two banks.

Cybersecurity firm CrowdStrike says the attacker used artificial intelligence tools, including models developed by China such as DeepSec, and even used the Cloud Code tool to find parties who could buy the stolen data.

AI company leaders anticipate that a catastrophic event would trigger a strong public and political backlash. They predict it would push an already skeptical public toward AI even more hostilely toward the technology and its leading figures, including Dario Amode, CEO of Anthropic, and Sam Altman, CEO of OpenAI.
OpenAI said its preparedness drills explore different scenarios without treating them as inevitable events, while Anthropic declined to comment.

political repercussions

Much of this secret preparation focuses on testing worst-case scenarios and preparing to brief the US Congress on the information before drafting anger-driven legislation.
It appears that the executives of these companies believe that the chances of enacting actual regulations are limited at the moment, but they nevertheless want to influence the rules that legislators may consider after a serious incident occurs.

The proposed responses range from halting the development of advanced artificial intelligence altogether to banning superintelligence completely.
Another proposal, which enjoys bipartisan support, would require advanced artificial intelligence systems to have a kill switch, although experts doubt that all AI systems can be realistically stopped in this way.

Behind all of this lie two deeper problems. The infrastructure for artificial intelligence has become so deeply embedded in the economy that imposing effective restrictions on it could lead to real economic disruptions.

At the same time, many AI models are already available for free download, which means that companies do not have much control over how these models are misused once they are made available to the world.

These preparations do not prove that a disaster is inevitable, but they do indicate that some AI leaders are taking this possibility so seriously that they want to equip policymakers before a real crisis forces action.


Post a Comment

0 Comments