OpenAI is considering the possibility of slowing down its advanced AI model development, according to a report by Bloomberg citing multiple sources on September 10.
Sam Altman, CEO of OpenAI, suggested during a company-wide meeting this week that the organization could adjust its AI development pace in coordination with other AI companies. However, he noted that there is a chance other companies may not agree to such a plan. Altman also mentioned that he discussed the need for speed regulation in AI development with White House officials in July.
This statement comes amid increasing competition among major AI companies, including OpenAI and Anthropic, to develop cutting-edge AI models, which has raised concerns about associated risks such as hacking. For instance, this week, Anthropic researcher Jacob Cox announced his resignation, warning that the competitive race in AI model development could lead to uncontrollable AI and potentially result in human extinction within a decade.
Additionally, at the end of July, over 1,100 employees from major AI companies, including OpenAI, Anthropic, Google, and Meta, issued a statement calling for the establishment of an international framework for regulating AI development speed. At that time, Jakub Pachocki, OpenAI's chief scientist, expressed concern, stating, "No one is prepared for the consequences of the rapid advancement of machine intelligence."
As the performance of AI models continues to improve, there are signs of malicious use. According to the New York Times, Anthropic reported that it blocked an attempt to use its AI model Claude for research that could lead to the development of biological weapons.
Meanwhile, as the risks associated with the AI model development race become more apparent, Jensen Huang, CEO of NVIDIA, predicted that the next major market for artificial intelligence will be in the cybersecurity sector. Speaking at a Goldman Sachs technology conference in San Francisco, Huang stated, "Cybersecurity will be the next major application area for AI," explaining that the automation of computer programming using AI models is accelerating the speed at which code can be exploited, and consequently, the speed at which code must be modified and secured is also increasing.
Altman also discussed measures to enhance power grid security against AI cyberattacks with U.S. utility companies this week, according to a report by Politico.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.